smith, hiawatha d., ph.d. digging deeper: understanding
TRANSCRIPT
SMITH, HIAWATHA D., Ph.D. Digging Deeper: Understanding Non-Proficient
Students Through an Understanding of Reading and Motivational Profiles. (2017).
Directed by Dr. Samuel Miller. 138 pp.
With the continued emphasis on accountability for students, schools are working
to increase the reading academic performance of their non-proficient students. Many
remedial approaches fail to identify the individual strengths and weaknesses and tend to
treat these students with a singular remedial focus on word identification (Allington,
2001). In this quantitative study, I explore the reading and motivational patterns present
with elementary non-proficient readers representing marginalized groups. The results
suggest that non-proficient readers do not need remediation with a singular focus, but
have unique needs that must be taken into account when planning remediation. This
study provided unique findings by examining the reading and motivational profiles of this
unique sample of students. Six profiles were identified that represented strengths and
weaknesses within the area of reading, as well as identifying preferred and less preferred
motivators. The study also supported the idea that motivation is multi-dimensional and
should be considered when providing support to struggling readers.
DIGGING DEEPER: UNDERSTANDING NON-PROFICIENT STUDENTS
THROUGH AN UNDERSTANDING OF READING
AND MOTIVATIONAL PROFILES
by
Hiawatha D. Smith
A Dissertation Submitted to
the Faculty of The Graduate School at
The University of North Carolina at Greensboro
in Partial Fulfillment
of the Requirements for the Degree
Doctor of Philosophy
Greensboro
2017
Approved by
______________________
Committee Chair
ii
ii
Pag
eii
2
2 P
ageii
22
APPROVAL PAGE
This dissertation written by HIAWATHA D. SMITH has been approved by the
following committee of the Faculty of the Graduate School at The University of North
Carolina at Greensboro.
Committee Chair ____________________________________
Samuel Miller
Committee Members ____________________________________
Terry Ackerman
____________________________________
Jewell Cooper
____________________________________
Melody Zoch
____________________________
Date of Acceptance by Committee
_________________________
Date of Final Oral Examination
iii
iii
Pag
eiii
2
2 P
ageii
i22
TABLE OF CONTENTS
Page
LIST OF TABLES ...............................................................................................................v
LIST OF FIGURES ......................................................................................................... vii
CHAPTER
I. INTRODUCTION ................................................................................................1
Problem Statement ......................................................................................4
Purpose of the Study ...................................................................................7
II. REVIEW OF LITERATURE .............................................................................10
Review of Reading Profiles Literature .....................................................12
Review of Motivation Profiles Literature .................................................24
Need for Marginalized Groups .................................................................31
Need for Economically Disadvantaged Students ......................................34
Recommendations .....................................................................................40
III. METHODOLOGY .............................................................................................44
Research Design ........................................................................................44
Data Analysis ............................................................................................57
IV. RESULTS ...........................................................................................................64
Reading Trends .........................................................................................65
Reading Factors ........................................................................................72
Reading Profiles ........................................................................................78
Motivation for Reading Trends .................................................................82
Motivation Factors ....................................................................................85
Motivation Clusters ...................................................................................95
Motivation and Reading Clusters ............................................................100
V. DISCUSSION ...................................................................................................106
Reading Profiles ......................................................................................107
Motivation Profiles .................................................................................108
Motivation and Reading Profiles ............................................................109
iv
iv
Pag
eiv
2
2 P
ageiv
22
Implications .............................................................................................114
Future Work ............................................................................................120
Limitations ..............................................................................................121
Conclusions .............................................................................................122
REFERENCES ...............................................................................................................124
APPENDIX A. MOTIVATION FOR READING QUESTIONNAIRE ITEMS ...........135
v
v
Pag
ev
2
2 P
agev
22
LIST OF TABLES
Page
Table 1. Comparison of Areas Assessed in Reading Literature .......................................22
Table 2. Instruments for Data Collection ..........................................................................56
Table 3. Descriptive Statistics for the Reading Data by School .......................................67
Table 4. MANOVA Test Results for Reading Differences Between Schools .................69
Table 5. Post Hoc Test Comparison Between Schools .....................................................69
Table 6. Descriptive Statistics for the Reading Data by Grade Level ..............................70
Table 7. Total Variance of Reading Factors .....................................................................74
Table 8. Reading Factor Communalities ...........................................................................75
Table 9. Reading Variables Factor Loadings ....................................................................76
Table 10. Reading Cluster History ....................................................................................80
Table 11. Reading Clusters Summary Data ......................................................................81
Table 12. Motivation Item Descriptive Statistics .............................................................84
Table 13. Initial MRQ Factor Structure by Instrument Authors .......................................87
Table 14. Revised Final MRQ Factor Structure ................................................................91
Table 15. Descriptive Statistics for the Motivation Data by School..................................93
Table 16. MANOVA Test Results for Motivation Differences Between Schools ...........94
Table 17. Descriptive Statistics for the Motivation Data by Grade Level ........................96
Table 18. MANOVA Test Results for Motivation Differences Between Grades .............97
Table 19. Motivation Cluster History ...............................................................................98
Table 20. Motivation Clusters Summary Data .................................................................99
vi
vi
Pag
evi
2
2 P
agev
i22
Table 21. Combined Cluster History ..............................................................................102
Table 22. Combined Clusters Summary Data .................................................................103
vii
vii
Pag
evii
2
2 P
agev
ii22
LIST OF FIGURES
Page
Figure 1. NAEP Trends Ethnicity Comparisons ................................................................34
Figure 2. NAEP Trends Economic Comparisons ..............................................................35
Figure 3. Scree Plot of the Reading Factors ......................................................................74
1
1
Pag
e1
2
2 P
age1
22
CHAPTER I
INTRODUCTION
Students’ academic performance continues to be a primary concern for our
nation’s schools and teachers. As teachers attempt to educate students, they face the
problem of ensuring that all students, including those from marginalized populations
(race, gender, religion, cultural group, or socioeconomic status) demonstrate proficiency
(Brown-Jeffy & Cooper, 2011; NCLB, 2001). As policy makers continue to implement
policies to address this goal, a focus is placed particularly on the performance of non-
proficient readers because all students are expected to read at grade level by the end of
the third grade. This focus is a result of federal accountability policies that evaluate
students’ performances on various accountability measures. This accountability places
emphasis on all students being successful with reading.
The National Assessment of Educational Progress Assessment is a nationally (bi-
annually) administered survey of achievement regarded as a source of information for
state-to-state comparisons as it assesses students’ ability to read within three different
contexts; reading for literary and informational purposes, and to perform a specific task
(schedules, directions, maps). NAEP scores have played a major role in the development
of educational policy over the last decade (Swanson & Barlage, 2006). The most recent
NAEP results (National Center for Education Statistics, 2015) showed that 64% of
fourth-grade students and 66% of eighth-grade students failed to meet its proficiency
2
2
Pag
e2
2
2 P
age2
22
level. The results showed a 1% point decrease (4th grade) and 2% increase (8th grade) in
the percent of student scoring below proficient on this assessment. Despite continued
efforts at school reform via mandates, over half of our students are still unsuccessful on
reading assessments.
As a result of nationwide testing via NAEP and other state level assessments,
additional mandates resulted in the creation of No Child Left Behind (NCLB), which
originally promised to have all students reading proficiently by the year 2014 (NCLB,
2001). The goals of NCLB emphasized proficiency of marginalized groups, including
economically disadvantaged, minority, and English Language Learner students. This
legislation required accountability systems for each state to annually monitor the
academic performance of students in grades 3-12. The primary signature of NCLB, thus
far, has been the imposition of high stakes testing of students, even though policy makers
know that children of poverty are more likely to fail than children of other socioeconomic
groups (McCaslin, 2009).
The Reading First initiative was developed as a part of NCLB and provided
funding for the identification of scientifically-based research to improve reading
instruction with the hope of reducing the number of non-proficient students (Learning
Point Associates, 2004). It provided four pillars to guide reading programs; valid and
reliable assessments, effective instructional programs and aligned materials, professional
development, and dynamic instructional leadership. The main result of this initiative was
the teaching of reading through phonics-driven instruction. This focus led to programs,
professional development, and materials with this stated isolated remedial purpose.
3
3
Pag
e3
2
2 P
age3
22
Several problems arose as a result of accountability legislation and subsequent
initiatives. The underlying problem is that the legislation’s emphasis on non-proficient
students resulted in them being treated as a homogenous group because their reading
scores fell within a similar performance range (Afferbach, 2004). If these students are
not homogenous, as Buly and Valencia (2002) claimed, then a one-size-fits-all emphasis
on remediation via the teaching of isolated word identification skills and a failure to
acknowledge other reading skills is misplaced. Previous researchers have warned not to
rely on a singular remedial focus via single assessments to make important instructional
decisions (Elmore, 2002; Shepard, 2000). These findings support the idea that additional
information about non-proficient students’ academic performance is necessary to help
them achieve success.
Despite these warnings, schools regularly address the needs of non-proficient
readers by providing intensive remedial instruction in various components of word
recognition (Allington, 2001). These remedial approaches are aligned with the common
assumption that phonics and phonemic awareness skills need to be mastered before
students can learn to comprehend texts (National Reading Panel, 2000). This response is
problematic in two ways. First, as stated earlier, non-proficient readers may not be a
homogeneous group: individuals within this grouping may have different instructional
needs than others. Second, this intervention only considers reading abilities without
attending to their motivations for becoming engaged in reading. This alternative is
particularly important because of recent studies showing the importance of students’
motivation for reading as a key component of their later reading success (Guthrie, Hoa,
4
4
Pag
e4
2
2 P
age4
22
Wigfield, Tonks, Humenick & Littles, 2007; Unrau & Schlackman, 2006; Wang &
Guthire, 2004).
Problem Statement
Schools’ reliance on a one-size-fits-all approach to assisting struggling readers
increased dramatically in recent years due to an intensified emphasis on test-driven
accountability (Allington & McGill-Franzen, 1992; No Child Left Behind, 2001; Slavin,
Cheun, Holmes, Madden, & Chamberlain, 2013). As stated earlier, remedial
interventions mainly focused on improving non-proficient readers’ word recognition
abilities (Allington, 2009; Fisher & Ivey, 2006). Questions have been raised about
whether such a one-dimensional focus adequately represents non-proficient students’
literacy needs (Afferbach, 2004; Allington, 2001; Buly & Valencia, 2002; Lin, 2000;
Shepard, 2000; Valencia & Buly, 2004). Such questions are important because
struggling students may need a variety of remedial approaches if they are to become
successful readers (Moje, 2004). As stated by Buly and Valencia (2002), one-size-fits-all
reasoning tends to treat the symptoms of not being a proficient reader without
understanding the many possible underlying causes for this lack of proficiency.
Buly and Valencia (2002) designed a study to identify these underlying causes,
inclusive of reading skills and strategies that characterize non-proficient readers’
strengths and weaknesses (Valencia, 2011). They examined the reading profiles of upper
grade elementary students deemed non-proficient on a state mandated reading assessment
because they failed to achieve at grade level. Instead of finding a common reason for
their struggles, the authors discovered ten profiles based on patterns of students’ ability to
5
5
Pag
e5
2
2 P
age5
22
identify words quickly (word identification), read fluently (fluency), and make meaning
from the text read (meaning). Their findings underscored the need to develop different
remedial programs based on students’ individual strengths and weaknesses. Their
research led to other studies where researchers identified multiple models of reading
profiles for struggling readers across grade levels (Dennis, 2013; Leach, Scarborough, &
Rescorla, 2003; Leseaux & Kieffer, 2010; Meyer, et al., 2013; Rupp & Leseaux, 2006;
Pierce, Katzir, Wold, & Noam, 2007).
The work of Buly and Valencia (2002, 2004) challenges the assumption of a one-
size-fits-all model for addressing the needs of struggling readers, and other researchers
have joined in the challenge with the creation of different profiles to describe non-
proficient students. While the identification of multiple profiles supports the need for
various remedial approaches, it also raises questions about the extent to which there are
consistent underlying structures represented in profiles across different studies. Without
an understanding of this underlying structure, a one-size-fits-all approach may be as
inappropriate for helping non-proficient readers as an any-profile-model approach. If we
are to support struggling readers, particularly those from marginalized student
populations, it is important to gain this understanding of the underlying constructs that
determine the successful engagement of non-proficient readers in their academic studies.
Reading researchers have shown that low performing students are not a
homogenous group (Dennis, 2013; Valencia & Buly, 2002); however, their profiles are
limited to cognitive domains. The continued barrage of assessments being forced on
students, as early as kindergarten, and focusing solely on isolated reading skills, continue
6
6
Pag
e6
2
2 P
age6
22
to impact their beliefs about reading and their reading performance. Previous researchers
have warned not to rely on a singular assessment to make important decisions about
children (Elmore, 2002; Shepard, 2000) and thus support the idea that additional
information about students’ values, beliefs and perceptions about reading are essential to
providing effective remedial support to assist them with achieving success.
Baker and Wigfield (1999) designed a study to identify motivational profiles.
They generated profiles of 5th and 6th grade regular education students. Through their
profiles, they found that children’s motivations for reading are multidimensional,
originating from varied backgrounds and experiences. They identified seven profiles
inclusive of different patterns of motivation from the areas of efficacy, importance,
curiosity, involvement, challenge, recognition, grades, competition, social, compliance
and work avoidance. Their profiles suggest that students should have differences in
motivation and these occur in varied ways and purposes. Therefore, no students should
be classified as motivated or not motivated, but should be assessed to determine their
motivational preferences for reading activities.
Though their work addresses the need for educators to understand the
motivational constructs represented by students, they failed to include three things. First,
despite the intense accountability focus placed on non-proficient students from
marginalized populations, their study did not specifically address these students. Second,
while they addressed motivation, they lacked an explicit connection to students’ reading
performances to better understand them holistically. Finally, the study examined ethnic
differences, but they did not include a specific focus on minority students because 52% of
7
7
Pag
e7
2
2 P
age7
22
their participants were Caucasian. This supports a need to specifically address
marginalized groups.
Purpose of the Study
Existing studies have addressed either the reading profiles of non-proficient
students or the motivational profiles of students in general, but the two have not been
bridged. As well, these studies have lacked a focus on understanding marginalized
student populations, including those from schools with high population of minorities,
particularly African American and Hispanic students. The purpose of this study is to
evaluate the reading and motivation profiles of elementary students, who have failed to
pass their state’s mandated reading assessment, and attend schools with similar student
demographics, including a large percentage of minority students.
Research Questions
This study examined the reading and motivational profiles of non-proficient
students across the upper elementary grades. The participating sample came from
schools with a high percentage of students representing marginalized groups, mainly
African-American, Hispanic, and those whose families lack economic resources.
1. What trends of reading and motivation for reading are represented in 3rd, 4th
and 5th grade non-proficient students?
2. What underlying motivational and reading constructs represent non-proficient
3rd, 4th and 5th grade students?
3. What are the reading profiles of 3rd, 4th and 5th grade non-proficient students?
8
8
Pag
e8
2
2 P
age8
22
4. What are the motivation profiles of 3rd, 4th and 5th grade non-proficient
students?
5. What are the reading and motivation profiles of 3rd, 4th and 5th grade non-
proficient students?
6. What do these profiles tell us about the instructional needs of 3rd, 4th and 5th
grade non-proficient students?
Significance of the Study
Existing studies have addressed the reading profiles of non-proficient students as
well as the motivational profiles of a generic sample of students, but the two have not
been connected. They have failed to place an emphasis on understanding marginalized
students, including students of color, especially those from low-income families, who
struggle with reading in the elementary grades. Studies have shown that struggling
readers can display a combination of both word recognition and meaning difficulties
(Buly & Valencia, 2002; Dennis 2012) and that students may have varying motivational
characteristics (Baker & Wigfield, 1999). When teachers provide instruction and
remediation to non-proficient readers, the work must include multiple areas of reading as
well as motivation. Motivation is a key factor in a student’s choice to read, their beliefs
about their reading, and the value they place upon the act of reading (Cambria & Guthrie,
2010; Schiefele, 1999). Through the use of motivation as a component in reading
profiles that directly address marginalized student groups, teachers might have a better
chance to increase the academic achievement of non-proficient minority students. In
addition, the profiles from this research can add to the existing literature by offering a
9
9
Pag
e9
2
2 P
age9
22
sample that comes directly from low income schools that are not identified as high
performing. Through this work, researchers and teacher educators can begin to think
about the specific reading factors represented as well as the role that motivation plays
specifically for non-proficient, low-income, minority students.
10
10
Pag
e10
2
2 P
age1
02
2
CHAPTER II
REVIEW OF LITERATURE
The primary signature of NCLB, thus far, has been the imposition of high stakes
testing of students to monitor academic performance; this monitoring has an equal goal
for all students, 100% proficiency, which has had negative implications for students’
performance and motivation throughout all schools (Au, 2009; McCaslin, 2009).
Increased accountability via testing was initially thought to be a positive achievement by
measuring and comparing standardized assessment results within schools and states, and
by motivating the unmotivated to learn (Orfield & Kornhaber, 2001); however, testing
has negatively impacted the motivation of students and placed limits on their
opportunities to reach academic expectations (Amrein & Berliner, 2003; Au, 2009).
An abundance of federal funds have been devoted to increasing achievement for
students. Unfortunately, despite the appearance of vast amounts of federal funding,
accountability-driven reforms with continued and frequent high-stakes testing only
confirm what we already knew regarding student reading achievement, in that, the same
populations continue to fail (National Center for Education Statistics, 2015). Two
possible reasons for continued failure include the following. First, there is a lack of any
emphasis on understanding students’ motivation. This includes acknowledging why
students become engaged in some activities and not others (Guthrie, Coddington, &
Wigfield, 2009) and using this information to modify instructional practices for students.
11
11
Pag
e11
2
2 P
age1
12
2
Second, the majority of the studies failed to include students from marginalized groups,
such as African American and Hispanic, particularly those who are economically
disadvantaged, the very students whom this legislation was intended to help. This lack of
inclusion of these groups limits our understanding of why these students have not
improved their achievement (Unrau & Schlackman, 2006).
The purpose of this section is to discuss and evaluate existing research related to
reading profiles and the need to address motivation profiles as well to more fully
understand why low-achievers continue to fail to perform at grade level. It will also
address the reasons and implications of addressing marginalized groups when attempting
to understand non-proficient readers. This section will be discussed in three major parts.
In the first part, I will discuss studies where researchers have identified profiles of readers
(non-proficient in isolation or with proficient students) to see if a common set of profiles
exists beneath different findings. I will do this by examining patterns of performance of
struggling students along with a highlight of the samples included. I will start this
evaluation with a more detailed explanation of the Buly and Valencia (2002) study,
followed by a review of subsequent studies of reading profiles, and then I will evaluate
the extent to which a common set of profiles or an underlying structure exists to help
educators understand the unique needs of struggling readers.
In the second part, I will evaluate whether the motivational profiles for non-
proficient readers represent the same variability as their reading profiles by examining the
patterns and constructs used in different studies. In this section, I will highlight existing
research that assessed the motivational patterns of students. This second part is important
12
12
Pag
e12
2
2 P
age1
22
2
because researchers have identified motivation as a critical factor in determining the
extent to which students become engaged in their studies (Baker & Wigfield, 1999;
Mazzoni, Gambrell, & Korkeamaki, 1999). The section will conclude with how my
suggested study fills an existing gap in the research.
In the third part, I will focus on whether certain marginalized groups have been
represented adequately within existing research. I will highlight the need to address these
groups in research related to reading and motivation profiles and how the implications of
the research will improve the outcomes for these groups of students. Last, I will finish
the review with recommendations for future research based on the gaps within existing
research.
Review of Reading Profiles Literature
In the initial study by Buly and Valencia (2002), their investigation did not unveil
a singular reason for failure on the mandated state proficiency test. Instead, the authors
used cluster analysis to group non-proficient students based on the similarities with their
reading performance levels. They discovered profiles based on students’ strengths and
weaknesses with identifying words quickly (word identification), reading fluently
(fluency), or making meaning from the text read (meaning). There were ten profiles
initially identified, grouped into four pairs, followed by two singular profiles. The
authors later (2004) combined the four pairs of double profiles based on their statistical
similarities for a total of six profiles to facilitate its understanding by practitioners.
Respectively, the grouped profiles included:
13
13
Pag
e13
2
2 P
age1
32
2
o Cluster 1 & 2: Automatic Word Callers 18%: The students in this cluster are
stronger in word identification and fluency than meaning.
o Cluster 3: Struggling Word Callers 15%: The students in this cluster are
experiencing some difficulty in word identification and meaning but able to read
with acceptable fluency.
o Cluster 4: Word Stumblers 18%: The students in this cluster exhibit strength in
making meaning but have difficulty with word identification and fluently reading
text.
o Cluster 5 & 6: Slow and Steady Comprehenders 24%: The students in this cluster
are lacking in fluency but their word identification and meaning abilities are
relatively strong, with meaning being the stronger.
o Cluster 7 & 8: Slow Word Callers 17%: The students in this cluster are displaying
difficulties in meaning and fluency with strength being in word identification.
o Cluster 9 & 10: Disabled Readers 9%: The students in this cluster are low in all
three areas and represent the smallest cluster of students.
In their reformulation, Valencia and Buly (2004) focused on the implications for
classroom practice for each of the profiles based on a prototypical student in each cluster.
Accordingly, the authors identified an underlying structure: there was only one cluster
with students performing low in all three areas, two clusters contained students with
strengths in more than one area (word identification and fluency, word identification and
meaning), and the remaining profiles consisted of strengths in only one area with the
other two areas as weaknesses. Each description provided an in-depth interpretation of
14
14
Pag
e14
2
2 P
age1
42
2
the profiles and their respective representations with regards to the literacy components.
Through their profiles, the researchers challenged whether the one-size- fits all phonics
and word-identification programs were appropriate for all students, a caution against
overgeneralizing students’ needs. As schools address the needs of non-proficient
students, they need to go beyond the scores of state-mandated standardized assessments
to identify student instructional needs.
The work of Buly and Valencia (2002, 2004) led to other studies where
researchers attempted to identify and analyze reading profiles, particularly for those
students who were identified as non-proficient (Dennis, 2013; Leach et. Al, 2003;
Leseaux & Kieffer, 2010; Meyer et. al, 2013; Rupp & Leseaux, 2006; Pierce et. al, 2007).
Using the work of Valencia and Buly (2002, 2004) as a template, I evaluated these
studies by examining the nature and scope of their measures, how they identified non-
proficient readers, and their analytical procedures. My intent was to determine the extent
to which profiles for struggling readers varied across studies and whether there was an
underlying structure beneath their profiles. Table 1 below presents a comparison of the
studies including the areas measured and assessments used. In this section I will review
each study, in chronological order, and then provide an evaluation of their common
properties and what they imply for the identification of reading profiles for struggling
readers. For each study, researchers used either clusters or profiles to identify their
categories. I will use the terms used by researchers when describing their results.
Leach, Scarborough, and Rescorla (2003) used eight measures of students’ word
recognition fluency, vocabulary, and comprehension abilities to identify the profiles for
15
15
Pag
e15
2
2 P
age1
52
2
3rd, 4th, and 5th grade students (n=161), from both affluent and economically diverse
populations, 5% of whom were minority. Splitting scores into high and low categories on
each of the measures, they identified four distinct profiles:
Profile 1: (8%) Comprehension deficit/No word deficit
Profile 2: (17%) Word deficit but no comprehension deficit
Profile 3: (16%) Deficit in both word and comprehension
Profile 4: (59%) No deficits in comprehension or word.
The profiles supported heterogeneity of reading development for these students with
those students identified as late-emerging reading disabilities being balanced in their
present within each profile.
Rupp and Leseaux (2006) investigated the profiles of proficient and non-
proficient 4th grade students (n=1,111). A lack of proficiency was determined based on
the students’ unsuccessful scores on district mandated reading assessments. They
assessed speed and accuracy for reading words; spelling; working-memory; and
phonological and syntactic awareness. The researchers used factor analysis to develop
high and low split scores for word-level skills, working-memory, and language skills,
then grouped students into four distinct clusters representing a combination of high and
low scores:
Cluster 1: (34%) Low Word, Low Memory
Cluster 2: (11%) Low word, high memory
Cluster 3: (16%) High word, Low memory
Cluster 4: (39%) High word, High memory.
16
16
Pag
e16
2
2 P
age1
62
2
Students classified as non-proficient were unequally represented in each cluster, with the
majority (76.8%) falling into cluster one.
Pierce, Katzir, Wold, and Noam (2007) evaluated urban 2nd and 3rd grade students
who attended an after-school program and scored more than two-thirds of a standard
deviation below the mean on one of the subtests or composite of the Test of Word
Reading Efficiency (TOWRE) assessment. Using factor scores for decoding, fluency,
text level skills, and vocabulary, they generated a four-cluster model using a cluster
analysis:
Cluster 1: (27%) represent students with high vocabulary and decoding scores,
low word and text level skill factors;
Cluster 2: (19%) represent students with high scores on word level efficiency, text
level, and vocabulary factors and average decoding scores;
Cluster 3: (28%) represent students with low scores on vocabulary and at or
above the mean level for decoding, word level, and text level;
Cluster 4: (26%) low in all four areas.
These profiles provided support for the idea that non-proficient readers have varied
reading strengths and weaknesses.
Leseaux and Kieffer (2010) identified the literacy profiles of students in 6th grade
who were language minorities and native English speakers in a low-income urban setting.
The Gates-MacGinitie Reading Comprehension assessment was used to identify students
who scored below the 35th percentile. Subsequent testing included vocabulary, decoding,
17
17
Pag
e17
2
2 P
age1
72
2
passage fluency, and working memory assessments. Using latent class analysis, three
profiles were generated:
Automatic Word Callers: (18.3%) these students are characterized by above-
average pseudo-word reading accuracy, substantially below-average vocabulary
skills, and average range fluency skills;
Slow Word Callers: (60.3%) these students are characterized by above-average
pseudo-word reading accuracy skills, far-below-average vocabulary skills, and
low-average fluency skills; and
Globally Impaired Readers: (21.4%) these students are characterized by below-
average performance on all measures.
An analysis of these profiles found that language status was not a predictor of
membership in the profiles.
Dennis (2013) evaluated 6th, 7th, and 8th grade students who failed the state
reading assessments the previous school year. The researcher included assessments for
phonemic awareness, phonics, fluency, vocabulary, and comprehension and identified the
study as a purposeful extension of the work of Valencia and Buly (2004) with a different
sample. Using cluster analyses, she identified four profiles using three factors which she
labelled as meaning, decoding, and rate:
Cluster 1: Slow and Steady Comprehenders (24%) These students have the
highest scores in meaning with low scores in decoding and rate;
Cluster 2: Slow Word Callers (26%) These students have highest scores in
decoding, with average meaning scores and low rate scores;
18
18
Pag
e18
2
2 P
age1
82
2
Cluster 3: Automatic Word Callers (24%) These students are highest in
decoding, with average rate scores and low meaning scores;
Cluster 4: Struggling Word Callers (26%) These students are highest in rate,
with average comprehension scores and low decoding scores.
These profiles demonstrate that non-proficient students have capabilities and areas that
require intervention.
Meyer et al. (2013) evaluated 5th and 6th grade students, who were identified as
non-proficient because they scored below the 50th percentile on mandated end-of-grade
standardized reading tests. The authors assessed word recognition in isolation, oral
reading accuracy and rate, vocabulary, and comprehension. They categorized students as
high or low on each assessment area (print processing and vocabulary) to develop four
profiles:
Cluster 1 (26%) These students have high scores on both print processing
measures and vocabulary measures;
Cluster 2 (48%) These students scored high on print processing measure but low
on vocabulary measures;
Cluster 3 (12%) These students scored low on print processing but high on
vocabulary measures; and
Cluster 4 (14%) These students scored low on print processing and vocabulary
measures.
Although the researchers assessed comprehension, it was not directly included as a part
of the profiles generated.
19
19
Pag
e19
2
2 P
age1
92
2
Before determining the extent to which each of these studies supports the
identification of a common set of profiles or an underlying structure, I examined each
study to see if the same dimensions of reading performance were evaluated. For this
purpose, I used the five components of reading, as identified by the National Reading
Panel (NICHD, 2000). Each of these components is interrelated, thereby raising the
possibility of students being strong or weak in one or more areas. While researchers have
criticized the Panel for its procedures and subsequent policies regarding literacy
instruction (Allington, 2009), there is general agreement as to the importance of their
components with promoting reading development. They include: (a) phonemic
awareness, (b) phonics, (c) fluency, (d) vocabulary, and (e) comprehension. A definition
of each component is presented below (Learning Point Associates, 2004):
A. Phonemic awareness —an awareness of and the ability to focus and
manipulate the individual sounds (phonemes) in spoken words. It includes
skills such as isolating phonemes, blending onset-rimes, blending and deleting
phonemes, adding and substituting phonemes, and segmenting words into
phonemes. This part of reading instruction allows students to understand
spoken words are made up of individual sounds.
B. Phonics —the study and use of sound/spelling relationships and syllable
patterns to help students read written words. It includes skills such as
identifying letters and sounds, and blending sounds. This is a part of reading
instruction that should not become a dominant component of a reading
program, but a means towards the end goal of reading, comprehension.
20
20
Pag
e20
2
2 P
age2
02
2
C. Fluency —reading text with sufficient speed, accuracy and expression to
support comprehension. This includes grouping words into phrases that are
easier to read. A lack of fluency requires a reader to use cognitive resources
for reading the words that could be used to make meaning from the text.
D. Vocabulary —the body of words and their meanings that students must know
and understand to comprehend text. This includes skills such as word parts or
roots and the use of context clues to gain the meaning of unknown words. In
order for a student’s vocabulary to increase, they must come in contact with
words outside his or her current vocabulary.
E. Comprehension —the ability to make meaning requiring specific skills and
strategies, vocabulary, background knowledge and verbal reasoning skills.
Comprehension includes strategies such as comprehension monitoring, asking
and answering questions, using prior knowledge, and summarizing what has
been read. This is the final goal of reading instruction.
Table 1 illustrates similarities and differences between the reading profile studies,
only Buly and Valencia (2002) and Dennis (2013) assessed each of the National Reading
Panels (NICHD, 2000) five components of reading in their identification of reading
profiles. Based on their factor analysis, phonemic awareness and phonics were included
under the label “word identification” (Buly & Valencia, 2002) and vocabulary and
comprehension were included under the label “meaning.” For both studies, fluency or
rate remained as an isolated factor. There are direct similarities between the strengths
and weaknesses identified in four of Buly’s and Valencia’s (2004) profiles and those
21
21
Pag
e21
2
2 P
age2
12
2
identified by Dennis (2013). The “Disabled Readers” group (9%) was not identified by
Dennis; quite possibly because she included older students, representative of multiple
grades, who probably did not have low scores on her word recognition measures because
such indices were designed primarily for younger students. For the profile, “Slow and
Steady Comprehenders”, it is possible that these students were placed in other categories
because of the nature of clustering process. Both Buly and Valencia (2004) and Dennis
(2013) used the cluster analysis methodology, but there was not a consistent description
across studies explaining the exact steps and criteria used to generate the profiles to
determine if this could be a cause for the difference in the number of profiles generated.
22
Pag
e22
2
2 P
age2
22
2
22
Table 1
Comparison of Areas Assessed in Reading Literature
Buly &
Valencia
(2002, 2004)
Leach,
Scarborough, &
Rescorla (2003)
Rupp &
Leseaux
(2006)
Pierce,
Katzir,
Wold, &
Noam
(2007)
Leseaux &
Kieffer (2010)
Dennis
(2013)
Meyer et al.
(2013)
Phonemic
Awareness
X X X
Phonics X Spelling X X X X
Fluency X X X X X X
Vocabulary X X X X X X
Comprehension X X Working
Memory
Text Skills Working
Memory
X
Number of
Profiles
10 then 6 4 4 4 3 4 4
23
23
Pag
e23
2
2 P
age2
32
2
Pag
e23
22
While these two studies failed to identify a similar number of profiles, they do
support the idea of a common underlying structure for non-proficient students. The most
common underlying structures were word recognition, fluency, and meaning
(comprehension and vocabulary). Based on the nature of the sample and analytical
procedures, researchers might discover additional profiles, particularly if the two
components of meaning – vocabulary and comprehension—formed separate factors. A
similar argument could be made if researchers evaluated the reading profiles of beginning
readers and the components of word recognition – phonemic awareness and decoding—
split into separate factors. Regardless, at this point, Buly and Valencia (2002) and
Dennis (2013) identified multiple profiles and their findings provided an underlying
structure by which educators could develop more multi-dimensional intervention
programs to meet non-proficient students reading needs.
Each of the remaining studies identified different profiles based on how the
authors assessed reading. For example, Leach, Scarborough, and Rescorla (2003) and
Rupp and Leseaux (2006) assessed fluency, but neither study found a separate factor for
this construct. Furthermore, studies by Pierce, Katzir, Wold, and Noam (2007); Leseaux
& Kieffer (2010); and Meyer et al. (2013) used different assessments for comprehension
and word recognition. Despite these differences in how constructs were assessed across
each of these studies, an underlying structure still appeared whereby non-proficient
students had difficulty with one of more of the identified areas.
After examining these studies, no one set of profiles represented all non-proficient
readers; there are multiple profiles with students’ performances varying across the
24
24
Pag
e24
2
2 P
age2
42
2
Pag
e24
22
dimensions of word recognition, meaning (vocabulary and comprehension), and fluency.
There is heterogeneity within the non-proficient student classification and this finding
requires the development of corresponding interventions to address the unique needs of
students. What still needs to be discovered, however, is whether different motivational
profiles exist among categories of non-proficient students, similar to what was discovered
in reading by Buly and Valencia (2002) and Dennis (2013). This link between
motivation and reading is important because numerous studies document the critical role
of motivation in understanding students’ academic performances (Mazzoni, Gambrell, &
Korkeamaki, 1999). The next section will be used to evaluate existing students focused
on motivation for reading.
Review of Motivation Profiles Literature
Consistent with the expectancy/value motivation theory, students’ willingness to
invest time and effort in academic studies depends on their expectations for success and
the perceived value of achievement (Atkinson & Feather, 1966; Eccles, J. S., et al., 1983;
Heckhausen, 1977). Researchers view motivation as complex and domain specific (Paris
& Turner, 1994; Wigfield, Guthrie, Tonks, & Perencevich, 2004) and a multifaceted
process, inclusive of choices and beliefs (Watkins & Coffee, 2004; Wigfield, Guthrie,
Tonks, & Perencevich, 2004). This process explains why students either approach or
avoid a task and the reasons for their engagement or lack thereof. Thus, on a daily basis,
motivation helps teachers to understand what attracts a student to start, continue, end, or
avoid an activity (Graham & Taylor, 2002). It is a key factor in understanding students’
choice to read, beliefs about reading, and the value they place upon the act of reading
25
25
Pag
e25
2
2 P
age2
52
2
Pag
e25
22
(Cambria & Guthrie, 2010; Eccles, Wigfield, & Schiefele, 1998; Schiefele, 1999).
Similar to reading profile research, teachers who employ a one-size-fits-all model to
promote motivation, quite possibly, fail to acknowledge its multifaceted nature, thereby
ignoring the individualized needs of some students (Valencia & Buly, 2004).
Motivation researchers have used the expectancy/value theory to conceptualize
approaches for understanding the non-cognitive factors impacting student achievement.
One popular approach focuses on learned helplessness, defined as a lack of persistence in
tasks that could realistically be mastered, usually because of a lack of effort caused by
repeated failures (Luchow, Crowl & Kahn, 1985). Such behaviors are problematic
because when students lack persistence, they give up, and thus have minimal chance for
success. Another example is the study of anxiety. People experience high levels of
anxiety when they believe that they are not competent to perform a certain behavior
(Stumpf, Brief, & Hartman, 1987), which interferes with their ability to attend, thereby
having a negative influence on their beliefs and efficacy for learning. Each example,
learned helplessness and anxiety show how students who lack expectancies for success
and do not value learning become alienated from their academic studies. This lack of
engagement leads to failure and eventually to being labeled as non-proficient. While
these approaches help educators to understand the behaviors and attitudes of non-
proficient students, they were not specific to any particular discipline; thus, we do not
know how these profiles apply to reading.
Wigfield and Guthrie (1995) used the expectancy-value theory to bridge the gap
between reading and motivation research (Atkinson & Feathers, 1966; Heckhausen,
26
26
Pag
e26
2
2 P
age2
62
2
Pag
e26
22
1977, 1991). They developed the Motivations for Reading Questionnaire (MRQ) to
define and evaluate students’ expectations and values regarding their motivation for
reading. Initially, this measure had 82 items, most of which were taken from Eccles’
Achievement Motivation Research Project (Eccles et. al, 1983) – ability and efficacy
beliefs, subjective task values, achievement goals, intrinsic motivation, along with items
related to attitudes about reading and motivation for reading.
In discussing motivation for reading, I will focus mainly on three studies. I will
summarize motivation prior to returning to these studies to look at how they attempted to
bridge the gap between motivation and reading. In the first study, Wigfield and Guthrie
(1997) studied 4th and 5th graders across two semesters and identified 53 items and 11
constructs using the MRQ. The constructs and their definitions are presented in Table 3
(Wigfield, 1997). Using factor analysis, the researchers determined the existence of three
higher order dimensions of motivation from these 11 constructs. The three dimensions
are Extrinsic Motivation (social, efficacy, involvement, curiosity, recognition, and
challenge), Intrinsic Motivation (Compliance, grades, recognition (Spring), and
importance), and Competition and Work Avoidance (Wigfield & Guthrie, 1997). The
dimensions’ structure was relatively stable across the two semesters.
In the second study, Baker and Wigfield (1999) extended the initial work of
Wigfield and Guthrie (1997) by directly examining links with motivation and reading
achievement, examining differences in motivation based on student characteristics, and
determining what motivational profiles exist for students by using data from the MRQ,
Reading Activity Inventory, Gates-MacGinitie Reading Test, Comprehensive Test of
27
27
Pag
e27
2
2 P
age2
72
2
Pag
e27
22
Basic Skills (Standardized Assessment) and a performance assessment. They used
confirmatory factor analysis of their 5th and 6th graders data to validate the identification
of the 11 constructs of Wigfield and Guthrie (1997).
Cluster analysis of the MRQ placed students into seven clusters (profiles):
1. Very Low Reading Motivation (n=14, 4%) The students in this cluster are
characterized by low scores across all constructs except work avoidance, in which
they scored the highest, just below the mean.
2. Low Reading Motivation (n=40, 11%) The students in this cluster had low ratings
across nine constructs with the exception of work avoidance, which were the
highest in this cluster. Their scores for competition were slightly below the mean.
3. Low Competition, Efficacy and Recognition (n=28, 8%) The students in this
cluster had the lowest scores in competition, efficacy and recognition and slightly
below average scores for compliance with the remaining constructs having
average scores.
4. Low Importance (n=28, 8%) The students in this cluster had average scores on
eight constructs. Importance scores were lowest, well below the mean, and
competition scores were slightly below the mean, with social scores falling just
above the mean.
5. Competitive and Work Avoidant (n=80, 21%) The students in this cluster have
average scores on eight constructs. Challenge scores were slightly below the
mean with competition and work avoidance scores falling just above the mean.
28
28
Pag
e28
2
2 P
age2
82
2
Pag
e28
22
6. Low Competition and Work Avoidance; High Importance and Compliance
(n=58, 15%) The students in this cluster had average scores on seven constructs.
Work avoidance and competition scores fell slightly below the mean with
compliance and importance scores slightly above the mean.
7. High Reading Motivation (n=123, 33%) This cluster contained the largest
percentage of students who had scores above the mean in all areas, with the
exception of work avoidance that fell slightly below the mean.
These clusters underscore the heterogeneity of students’ motivation for reading as
students were somewhat evenly split among the clusters, with no cluster containing more
than 33% of the sample. Students in the first two clusters, labeled very low and low
reading motivation were characterized by high work avoidant scores. Students in the last
two clusters Low Competition and Work Avoidance and High Reading Motivation were
the opposite with low work avoidance scores. The remaining clusters varied by levels of
competition, need for recognition, importance attached to reading, and efficacy. Only in
one cluster, low competition and work avoidance: high importance and compliance, did
students demonstrate high scores on all the positive reading indices.
In the third study, Guthrie, Coddington, and Wigfield (2009) determined that
existing motivation literature lacked a specific focus on the constructs (intrinsic
motivation, avoidance, self-efficacy and perceived difficulty) that contribute to positive
motivation or undermine motivation for reading. With this in mind, they wanted to
capture and better understand the relationships between these constructs. They evaluated
5th grade African American and Caucasian student responses for only two of the four
29
29
Pag
e29
2
2 P
age2
92
2
Pag
e29
22
motivation dimensions: intrinsic motivation and work avoidance to identify four profiles.
Using a different analytical approach, the researchers ordered students’ scores from
highest to lowest, then separated students into distinct groups of high or low by splitting
the scores at the median. This approach allowed the researchers to form profiles
consisting of clear independent constructs. They identified four profiles. They included:
1. Avid (high intrinsic and low avoidance) students who have reading interests,
enjoy reading in and out of school, and do not avoid school reading.
2. Apathetic (low intrinsic and low avoidance) students who are low on intrinsic
reading and avoidance of reading.
3. Ambivalent (high on intrinsic and high avoidance) students who have intrinsic
motivation for some texts but not others with avoidance of reading high for
some kinds of reading.
4. Averse (low intrinsic and high avoidance) students who are actively opposed
to most kinds of reading and possess few reading interests.
The methodology used allowed for clearly defined profiles. Each profile contains
students with either high or low scores on each of the two constructs.
Across the three studies, given the consistency with some constructs and the
stability of their findings at different points of time, these results underscored the
potential for the MRQ to develop specific profiles for different types of non-proficient
readers. The one challenge to developing these profiles relates to the types of reading
measures included or not included in these studies.
30
30
Pag
e30
2
2 P
age3
02
2
Pag
e30
22
In the first study, Wigfield and Guthrie (1997) used the Reading Activity
Inventory, a measure of the breadth and depth of students’ personal reading habits as
their reading component. They used this instrument because such behaviors were a
strong predictor of reading achievement. They found the social, self-efficacy, curiosity,
involvement, recognition, grades, and importance constructs had the strongest
relationship to reading activity (Wigfield & Guthrie, 1997). Because this measure looked
at personal reading habits instead of actual classroom performances, it is of limited use
for identifying different profiles in the classroom for non-proficient readers.
In the second study, Baker and Wigfield (1999) used the Reading Activity
Inventory, the Gates-MacGinitie Reading Test, the Comprehensive Test of Basic Skills
(Standardized Assessment) and a researcher-developed performance assessment, where
students read passages and answered open-ended questions to measure reading
performance. With personalized reading habits, the results were obvious – students who
had the greatest breadth and depth with their reading habits had more positive
motivational outcomes than did students who read less widely or frequently. With the
formal reading measures, no differences were found between the clusters. Finally, with
the performance measure, only one difference was discovered: students in the sixth
profile significantly outscored students in the other profiles.
In the third study, Guthrie, Coddington, and Wigfield (2009) used the Gates-
Macginitie Reading Test, Woodcock-Johnson Fluency Test, and researcher-developed
word recognition test to measure reading performance. The only finding was a positive
relationship between scores on the Gates-McGinitie and intrinsic motivation for the
31
31
Pag
e31
2
2 P
age3
12
2
Pag
e31
22
Caucasian students. No such relationship was found for the African-American/Black
students.
The existing research has supported the need to understand non-proficient
students from both reading and motivation for reading perspectives. What is missing is
an understanding of which non-proficient students should be the focus of research. The
next section will examine the research to determine which groups of students are not
adequately represented and why they should be the focus of future research.
Need for Marginalized Groups
While there is evidence of a need to understand non-proficient students from both
a cognitive and non-cognitive perspective, there is a need to understand specific groups
of these students, particularly those students who have not performed well despite
schools’ efforts to address their needs. Au (2009) referred to students, primarily minority
students with families who lack economic resources, as residing in “zipcode” schools
because their geography was a better predictor of their overall achievements in public
schools than any other measure of their performance. One possible explanation for this
effect is the nature of our reforms to improve their performances, in that, students might
need interventions that go beyond the present focus on improving word recognition
abilities. As a result, I argue that we need accurate evaluations of their reading strengths
and weaknesses if we are to break this cycle of underachievement for these student
populations.
Since the Coleman Report in 1966, a distinguishable performance gap has been
identified between Caucasian and African American students has been a concern for
32
32
Pag
e32
2
2 P
age3
22
2
Pag
e32
22
educators, researchers, and policy makers (Clotfelter, Ladd, & Vigdor, 2009). While the
term achievement gap carries many negative connotations, this term is used as this term
was used by researchers to support the gap. The preferred term would be opportunity gap
as this removes the deficit perspective present within the other term. As well, the
achievement gap has been noted between students from low and high socioeconomic
backgrounds (Milner, 2013). This achievement gap is wide and despite numerous efforts
and initiatives, continues to exist (Lee, Grigg, & Donahue, 2007). NCLB placed a strong
focus on the differences in achievement by identifying equal performance requirements
for each group of students that must be met each year through minimal growth
requirements. To close the performance gap for minority students, schools must provide
high quality instruction (Au, 2009) and understand the performances represented in each
group. Without a direct focus on understanding what reading and motivational constructs
are represented in these groups, there will continue to be interventions and strategies that
fail to address the needs of students from these marginalized groups. To understand the
need for representation of certain students, a few questions must be answered. Is there an
opportunity (achievement) gap? What specific groups are those which are truly a part of
this gap?
The most recent NAEP assessment results (National Center for Education
Statistics, 2015) showed a 1% point decrease (4th grade) and 2% increase (8th grade) in
the percent of student scoring below proficient on this assessment since the previous
assessment. Despite continued efforts at school reform via mandates, over half of our
students are still unsuccessful on reading assessments. Although we see a lack of success
33
33
Pag
e33
2
2 P
age3
32
2
Pag
e33
22
overall, there are differences in achievement for many of the groups who are identified
through the NCLB legislation as marginalized groups. To visually represent the trends of
performance and the gaps that exist between the marginalized and non-marginalized
groups, I created the line graphs representing differences in proficiency for minority and
non-minority groups. These graphs represent the fourth-grade reading proficiency
performance on the NAEP from 2002, the first assessment prior to NCLB, through 2015,
the latest administration of the NAEP (National Center for Education Statistics, 2015).
The first graph represents the performance trends for Caucasian, African American, and
Hispanic students.
Figure 1 illustrates several key accountability patterns related to ethnicity. First,
there is an increase in the performance for the three groups from 2003 through the last
administration in 2015. Second, while there has been an increase in performance for
these three groups, the achievement gap between Caucasian students and ethnic minority
students continues to exist. The achievement gap since 2002 between Caucasian and the
ethnic minority groups has decreased minimally, but at no point did the performance
level for African American or Hispanic students meet the baseline of Caucasian students
after implementation of NCLB in 2002, Last, between the previous (2013) and current
administration (2015) of the NAEP, there was no growth noted for the Caucasian nor
African American groups, with a single percentage point increase for Hispanic students.
34
34
Pag
e34
2
2 P
age3
42
2
Pag
e34
22
Figure 1. NAEP Trends Ethnicity Comparisons
Need for Economically Disadvantaged Students
Similar to the first graph, Figure 2 identifies some key accountability patterns
related to economic status. First, there has been significant positive growth since 2003
for both economically advantaged and economically disadvantaged students. Second,
while there has been growth noted, there is still a significant achievement gap between
students representing these two economic categories. Students who are from
economically disadvantaged situations continue to perform at lower levels despite
continuous legislation. At no point did students in the lower economic group match the
baseline performance of the higher economic group. Last, although there is growth,
students from economically disadvantaged situations are not growing at a rate
comparable to their counterparts. Economically disadvantaged students have grown a
total of 6% in the 12 years since NCLB, while their counterparts have grown 10%.
35
35
Pag
e35
2
2 P
age3
52
2
Pag
e35
22
The achievement gap from assessment data is a reality for both economically
disadvantaged and African American students. Despite continuous reforms and
legislation, the gap does not appear to be closing, but is larger than when initially
assessed via NAEP reading assessments. Therefore, it is essential to include these
marginalized populations in research to better understand their patterns of cognitive
and/or non-cognitive factors. It is essential because policy effects and responses must be
carefully studied to understand the impact on closing achievement gaps for marginalized
groups.
Figure 2. NAEP Trends Economic Comparisons
36
36
Pag
e36
2
2 P
age3
62
2
Pag
e36
22
Are These Groups Represented?
I will use this next section to evaluate some of the existing literature on non-
proficient readers and motivation for reading to determine if marginalized minority
students such as African American, Hispanic, and Economically Disadvantaged students
were adequately represented. Although NCLB provides a direct focus on several
marginalized groups, for purposes of this study, I will place emphasis on the
representation of Ethnic Minorities and Economically Disadvantage students as these
groups have shown historically wide achievement gaps with their counterparts (see
NAEP graphs above).
Buly and Valencia’s (2002) work has served as a guide for many later research
studies using reading profiles to understand non-proficient students. Neither of their
publications adequately identified those characteristics of their students to determine if
they included the previously stated populations. The sample was taken from one school
district from the northwestern part of the U.S. with 57% of the students Caucasian and
43% noted as students of color. From this 43%, 11% were Hispanic and 11% were
African-American. Last, 47% of the students were considered economically
disadvantaged. While this study has great implications for research and practice, it
described the district’s characteristics and not the sample’s.
Leach, Scarborough, and Rescorla (2003) used eight literacy measures of
students’ word recognition including fluency, vocabulary, and comprehension abilities
for 3rd, 4th, and 5th grade students (n=161) with reading difficulties. The researchers split
students into groups with high and low combinations of word recognition and
37
37
Pag
e37
2
2 P
age3
72
2
Pag
e37
22
comprehension performance to identify four profiles representing the patterns of these
students. While their study had a unique sample in the inclusion of students with
exceptional needs, their study did not adequately address other marginalized groups. Of
the 161 students representing 12 elementary schools, only half of the schools had less
than 9% of economically disadvantaged students. The other half ranged from 12 to 60%
of students from economically disadvantaged households. Of the sample, 95% of the
students were Caucasian, and the remaining percentage was a mix of African American,
Asian, and Hispanic. When looking at marginalized groups identified from NCLB, the
study did address students with special needs; however, it did not adequately represent
the patterns of performance for students who are minorities nor those who are
economically disadvantaged.
Pierce, Katzir, Wolf, and Noam (2007) evaluated urban 2nd and 3rd grade students
who scored more than two-thirds of a standard deviation below the mean on one of the
subtests and/or composites of the Test of Word Reading Efficiency (TOWRE)
assessment. They used factor analysis and a high/low split of scores to form their four
profiles. The 140 students in their study were from five schools between Phoenix and
Boston. The sample was composed of 60% Hispanic and 12% African American
students. There is not an adequate representation for African American students.
Additionally, while researchers described the income levels of students by sharing the
percentage of mothers and fathers who earned less than $35,000 a year, it was not linked
to its specific populations. To interpret poverty level, readers would need to know the
family size as well as specific amount of money each parent earns, therefore this
38
38
Pag
e38
2
2 P
age3
82
2
Pag
e38
22
information will not be considered to adequately represent the economically
disadvantaged NCLB group.
Leseaux and Kieffer (2010) identified the literacy profiles of students in 6th grade
who were language minorities and native English speakers in a low-income urban setting.
Their sample of 262 represented students from five middle schools and one elementary
school where 201 were language minority students (English as a Second Language) and
61 were native English speakers. The six schools had low-income populations ranging
from 44% to 100%, and 10% of the sample had special education designations. When
thinking about marginalized groups, this study adequately addresses the English
Language Learners (ELLs) population; however, we are unaware of the numbers of
ethnic groups. The inclusion of students from economically disadvantaged populations
was difficult to infer because it is not specifically identified.
Dennis (2013) evaluated 6th, 7th, and 8th grade non-proficient students who failed
the state reading assessments the previous school year. The sample for the study
included 94 middle school students from four middle schools. Of those 94, 56% were
Caucasian, 36% were African American, and 7% Hispanic. The percentage of students
from economically disadvantaged households was 82% with 36% receiving special
education services and 10% being classified as English learners. The representation of
African American students was marginal, which did not provide adequate representation
to understand the reading constructs present within the group. The sample does
adequately represent students from economically disadvantaged groups. With 82%
39
39
Pag
e39
2
2 P
age3
92
2
Pag
e39
22
represented, these findings do have some generalizability to other students from
economically disadvantaged groups.
Meyer et al. (2013) evaluated 5th and 6th grade students who were identified as
non-proficient on a mandated end-of-grade standardized reading tests. The sample was
barely described in detail in that 65 students were from a rural, small school system with
a total of 1,800 K-12 students. No information related to ethnicity or socio-economic
status was included.
The motivation studies did not provide adequate descriptions of the sample.
Baker and Wigfield (1999) included a sample of 371 students from five schools from a
large mid-Atlantic U.S. city. They did not have ethnicity and income level information
for 75 students due to a school removing itself prior to the final data collection. For the
sample information known, 52% were Caucasian, 46% were African American, and 54%
were economically disadvantaged. Additionally, they did not explain the percentage of
economically disadvantaged within the sample. Guthrie, Coddington, and Wigfield
(2009) included 245 5th grade students from three schools in a mid-Atlantic state. The
sample was 76% Caucasian, 24% African American, and 10% received special education
services. No information was reported to describe the income status of the participants.
Wigfield and Guthrie (1997) sample was 70% Caucasian and 30% African American
students and the income status was not reported.
While each of these studies added pertinent information to the literature in their
respective areas, do they adequately represent the marginalized groups who have been
identified as non-proficient students? The answer is somewhat. There were positive and
40
40
Pag
e40
2
2 P
age4
02
2
Pag
e40
22
negative implications from existing literature. First, the studies have some adequate
representation of students from some marginalized groups, including special needs,
Hispanic students, ELLs and economically disadvantage students. While these groups
were represented, they were inconsistent in their representation. Second, within several
studies, there was a failure to identify the specific sample or to make a general reference
to the population. This limits knowledge of who the actual sample included. Next, with
the exception of one study, the others all included samples representative of three or more
schools (with some of these schools not having high percentages of minority or
economically disadvantaged populations). The schools included lacked a high minority
or low-income population. Last, with all of the studies, there was not an adequate
representation of African American students.
Recommendations
While these studies supported the notion of non-proficient students in reading
demonstrating variability in their motivational profiles, there are several caveats
regarding what needs to be accomplished in order to more accurately identify the
underlying dimensions of non-proficient readers’ motivational profiles. The following
recommendations are related to identification of these inclusive profiles.
First, based on its use in existing research studies, the MRQ is the best measure to
use when assessing the motivational profiles of non-proficient elementary grade students.
With the complexity of motivation, it needs to be assessed with instruments that are
specific to the content area of reading and age/grade appropriate. The MRQ encompasses
a holistic representation of a student’s motivational dimensions for reading through its
41
41
Pag
e41
2
2 P
age4
12
2
Pag
e41
22
comprehensive measures. With the strong prevalence of the MRQ for developing an
understanding of students’ motivation for reading, it can serve as the foundation for
understanding the motivational components of non-proficient students.
Other researchers (e.g., Chapman & Tunmer, 1995; Gambrell, Palmer, Codling, &
Mazzoni, 1996; McKenna, M.C., Kear, D.J., & Ellsworth, R. A., 1995) developed similar
instruments to assess motivational constructs. Each of these instruments has smaller
subset scales that are directly related to those of Wigfield and Baker (1997). The MRQ
has been utilized in research on reading motivation with students in upper elementary
grades and early adolescent age to examine and determine the dimensions that exist for
children’s motivations for reading. With the inclusion of these areas, it is the most
relevant to use with developing a better understanding of students reading values, beliefs,
and behaviors in grades 3-5.
Second, similar to reading profile research, what is important is not the number of
profiles but the underlying structures of constructs. There is a need to include all the
items on the MRQ to allow underlying motivational dimensions to be revealed through
the appropriate methodology. Therefore, I believe a confirmatory factor analysis should
be used with the MRQ and exploratory factor analysis with reading assessments, similar
to what has been done by previous researchers. Using factor analysis, Wigfield and
Guthrie (1997) confirmed the existence of their suggested 11 motivation for reading
domains with their instrument; while Buly and Valencia (2002) and Dennis (2013)
identified underlying constructs of meaning, fluency, and decoding. While previous
researchers identified these constructs and dimensions, are they consistent across multiple
42
42
Pag
e42
2
2 P
age4
22
2
Pag
e42
22
samples? Watkins and Coffey (2004) found that there is variability within the way the
motivational constructs are represented in students. Along these lines, the same
variability could exist for the reading factors.
Third, a specific recommendation based on existing motivational profiles is that
profiles be examined that specifically reflect the motivational values of non-proficient
students. The existing profiles focus on ethnic or economic basis groups without a focus
on non-proficient students as a special group. Future research should identify students
who are non-proficient with a goal of understanding the relationship between their
motivation and literacy profiles.
Fourth, in future studies, certain elements should be included related to non-
proficient students. These elements include addressing students from economically
disadvantaged homes, placing an emphasis on ethnic minorities. The current studies seek
to identify causes of disparities for non-proficient students, but in truly addressing the
demands of NCLB and other accountability legislation, there should be inclusion of
students from economically disadvantaged homes, placing an emphasis on ethnic
minority students, and including students, who are from schools with high percentages of
economically disadvantaged students. The achievement gap is impacted by school
experiences and research findings (Burchinal et al., 2011); therefore, researchers must
include students from these specific marginalized populations with historic achievement
gaps.
The final recommendation is to link motivation with reading. Without this
connection, educators will not understand how these two important areas are linked. A
43
43
Pag
e43
2
2 P
age4
32
2
Pag
e43
22
fuller understanding of this connection will help educators to differentiate their
instruction.
44
44
Pag
e44
2
2 P
age4
42
2
Pag
e44
22
CHAPTER III
METHODOLOGY
Research Design
The goal of the study was to identify reading and motivational profiles of non-
proficient marginalized students in zipcode schools, where schools within geographical
areas tend to represent a specific socio-economic status and ethnic makeup. With this
goal, I planned to discover different patterns of reading and motivation for reading of this
sample of students. The research questions suggested a quantitative study using several
multivariate analysis techniques to generate and analyze the reading and motivational
profiles of non-proficient students across the upper elementary grades. The research
questions that guided this study were as follows:
1. What trends of reading and motivation for reading are represented in 3rd, 4th
and 5th grade non-proficient students?
2. What underlying motivational and reading constructs represent non-proficient
3rd, 4th and 5th grade students?
3. What are the reading profiles of 3rd, 4th and 5th grade non-proficient students?
4. What are the motivation profiles of 3rd, 4th and 5th grade non-proficient
students?
5. What are the reading and motivation profiles of 3rd, 4th and 5th grade non-
proficient students?
45
45
Pag
e45
2
2 P
age4
52
2
Pag
e45
22
6. What do these profiles tell us about the instructional needs of 3rd, 4th and 5th
grade non-proficient students?
Participants
The results of accountability legislation continue to directly impact students in
“zipcode” schools with large numbers of marginalized students who are living in poverty
and considered ethnic minorities (Gaddis & Lauen, 2014; Hursh, 2007; Jackson, Johnson,
& Persico, 2010). While these groups should be the focus of much research with reading
profiles, analysis of previous research demonstrated a lack of adequate representation of
these groups. As such, these schools and these students must be represented in research
to show the patterns that represent their reading and motivations for reading. Therefore,
my sample for this dissertation was purposefully representative of several of these
populations. For purposes of this study, non-proficient 3rd, 4th, and 5th grade students are
described as those who received a score of level 1 or 2 (out of 5) on the English
Language Arts End of Grade assessment for the previous academic year.
After university and district approval, contact was made with the principal and
administrative team of the identified schools. To follow FERPA and district regulations,
the schools were provided with recruitment letters and permission forms to share with
students who were eligible for the study (Scoring level 1 or 2 on the English Language
Arts End of Grade assessment). School administration determined eligible students based
on the non-proficient criteria and shared the forms with parents of those students as part
of the recruitment process. Upon return of the permission slips, students began
participation in the study.
46
46
Pag
e46
2
2 P
age4
62
2
Pag
e46
22
The students were assessed by the author and his doctoral supervisor. Students
were assessed during times designated by the administration that limited distractions
from instructional time, including their intervention/enrichment and
specials/enhancements times. All measures were administered to students in three to four
sessions. During session one, students signed the assent form, then completed the
CTOPP, QRI word list, and the PPVT assessments. In the second assessment, students
completed the Narrative and Expository QRI’s. In the third session, in a group session, I
administered the Motivation for Reading Questionnaire to students. The maximum time
for assessment completion was two hours with data collection occurring between January
and March, 2017.
The research study was conducted in schools in two neighboring school districts
in North Carolina. Three schools were selected for participation to acquire as many
participants as possible that fit the desired criteria of “zipcode” schools and a high
percentage of minority and low socioeconomic status populations. The schools are
within a close proximity to each other. All students in each school received free or
reduce breakfast and lunch. One school was selected from District A and two from
District B. District A is a large school district that includes 40.62% Black/African-
American, 33.4% Caucasian, 15.2% Hispanic, 6.3% Asian, and 4.5% other race/ethnicity.
District B is smaller with 62.5% White, 20.2% Black/African-American, 11.4% Hispanic,
and 5.9% other race/ethnicity.
Demographic data summarizing the study were generated using SAS software
(SAS Institute, 2013). In District A, School 1 had a student population of 569 students---
47
47
Pag
e47
2
2 P
age4
72
2
Pag
e47
22
67% Black, 17% Hispanic, 12% white and 4% other, with all students receiving free
lunch. The sample is representative of the overall school population with the majority of
the students identified as Black (81%), with the remaining 19% being made up equally
(6.3%) of White, Hispanic, and Multi-racial/Other students. The sample contained an
unequal ratio of male (40.6%) to female (59.4%) students. When examining the grade
and age frequencies, almost half of the sample consisted of third graders (46.9%) with
fourth (31.3%) and fifth grades (21.9%) representing the rest of the sample.
In District B, the superintendent selected two of the districts lower performing
schools that were both in close proximity to each other and also included high
percentages of minority non-proficient students. School 2 has a student population of 540
students---39.3% White, 31.5% Hispanic, 22.5% Black/African-American, and 6.7%
Multi-racial/Other, with all students receiving free lunch. The sample is similar to that of
the overall school population, the majority of the students (68.9%) identified as Hispanic
(36.4%) or White (32.5%) with the remaining 31.1% being similar with Multi-
racial/Other (16.9%) and African-American/Black (14.3%). The sample contained an
unequal ratio of male (50.6%) to female (49.4%) students.
School 3 has a student population of 417 students--57.1% Black/African
American, 11.3% Hispanic, 22.3% White, and 9.3% Multi-racial/Other, with all students
receiving free lunch. The sample is representative of the school population with the
majority of the students identified as Black (63%), with the remaining 37% being made
up of two of the other major ethnic group: White (28.3%) and Hispanic (8.7%). The
sample contained an unequal ratio of male (39.1%) to female (60.9%) students.
48
48
Pag
e48
2
2 P
age4
82
2
Pag
e48
22
The combined sample included 187 non-proficient readers. The majority of the
sample were considered ethnic minorities (77.5%). Approximately half of the sample
were African-American/Black students (49.2%). The remaining half consisted of the
other three groups: White (22.5%), Hispanic (19.3%) and Other (9.1%). The sample
contained an unequal ratio of male (44.4%) to female (55.6%) students.
When examining the grade frequencies, almost half of the sample consisted of
third graders (44.4%), with fourth (28.9%) and fifth grades (26.7%), almost equally
representing the rest of the sample. The frequencies demonstrated the uniqueness and
importance of the sample in comparison to samples used within existing studies in that
this population primarily includes students of color from families with minimal economic
resources. A sample of this type is valuable as it is absent from existing research.
Measures
Reading is a complex process. Acknowledging this complexity, researchers have
assessed a variety of reading components, using a variety of assessments, to better
understand the patterns of students’ reading performance. With the need to focus on non-
proficient students, specifically ethnic and economic minorities, it is my belief that
assessments should be reflective of the five components of reading, as identified by the
National Reading Panel (NICHD, 2000). Each of these components is interrelated,
thereby raising the possibility of students being strong or weak in one or more areas.
While researchers have criticized the Panel for its procedures and subsequent policies
regarding literacy instruction (Allington, 2009), there is general agreement amongst
literacy researchers as to their importance with promoting reading development. They
49
49
Pag
e49
2
2 P
age4
92
2
Pag
e49
22
include: (a) phonemic awareness, (b) phonics, (c) fluency, (d) vocabulary, and (e)
comprehension. Each component is described below (Learning Point Associates, 2004).
As identified in Chapter 2, only the work of Buly and Valencia (2002) and Dennis
(2013) assessed all five components in the creation of their reading profiles. An analysis
of their samples revealed neither of their samples adequately addressed African American
students. However, Dennis’ sample did include students identified as economically
disadvantaged. Therefore, to extend their work, I addressed the missing diversity within
the existing samples by including larger numbers of ethnic minority students (>50% of
the sample) and those students representing economically disadvantaged backgrounds.
Similar to the analysis of reading work, the existing motivational work fails to
adequately address African American and economically disadvantaged groups. In the
study by Baker and Wigfield (1999), we are not knowledgeable of the actual true sample
with such a large amount of missing demographic information. Although there was some
representation of these marginalized groups, it was not an adequate representation.
With the complexity of motivation, it needs to be assessed with instruments that
are specific to the content area of reading and age/grade appropriate. The MRQ
encompasses a holistic representation of a student’s motivational dimensions for reading
through its comprehensive measures. With the strong prevalence of the MRQ for
developing an understanding of students’ motivation for reading, it can serve as the
foundation for understanding the motivational components of non-proficient students.
Based on its use in existing research studies (Baker & Wigfield, 1999; Guthrie, et al.,
2007; Unrau & Schlackman, 2006; Wang & Guthire, 2004), the MRQ is the best measure
50
50
Pag
e50
2
2 P
age5
02
2
Pag
e50
22
to use when assessing the motivational profiles of non-proficient elementary grade
students. The MRQ has been utilized in existing motivation research; however, the
instrument needs to be used with different samples to determine which of the suggested
domains are consistently represented (Yuan, 2005). Although the instrument has
theoretical support, the presence of each domain was validated using statistical analysis
to determine which were represented with this unique population of students.
In the next sections, I will present information on the suggested instruments to use
when collecting data to generate the reading and motivational profiles. A table is
included at the end of this section to summarize each area that will be assessed, the
assessments used, and the scores generated from the assessments (see Table 2). I will
start with summary information on each assessment. This will include an overview of the
reading construct(s) assessed by the instrument, how the assessment is administered, and
reliability and validity information for the instrument (if available). The assessments will
be summarized starting with the lower level reading skills moving to higher levels, then
finishing with motivation.
Phonemic Awareness
Phonemic awareness is the ability to recognize, identify, and manipulate sounds
in words. The Elision Subtest of the Comprehensive Test of Phonological Processing
(CTOPP) was used to measure phonemic awareness as it assesses the advanced levels of
the phonemic awareness developmental spectrum. This test required students to isolate
or remove syllables or phonemes within 20 spoken words (Wagner, Torgeson &
Rashotte, 1999). Although traditionally administered to student’s ages 5 and 6, the
51
51
Pag
e51
2
2 P
age5
12
2
Pag
e51
22
Elision subtests have a form for older students allowing it to be administered to ages 7-
24. Test Reliabilities for the assessment range from .74 to .97 for the subtests with
internal consistency coefficients for all composites of the assessment were all .85 and
above.
Comprehension
The Qualitative Reading Inventory-4 (QRI) (Leslie & Caldwell, 2006) is a
commercial informal reading inventory that determines narrative and expository reading
levels for students. The QRI-4 can be used for multiple purposes. For this study, it was
used to assess fluency (reading text with sufficient speed to support comprehension), as a
measure of students’ rate in word identification (word list and accuracy), and as a
measure of comprehension (ability to make meaning from the text read). The fluency
and accuracy data were used to determine proficiency for each grade level text read,
therefore; for a student to achieve a grade level, they must have had sufficient accuracy
and fluency on that level text. For this dissertation, only the word list grade equivalent
scores and the highest instructional comprehension level grade equivalent scores for
expository and narrative texts were reported and analyzed.
The appropriate starting level for narrative text was determined by having
students read aloud word lists representing grade level words until they are no longer able
to read with at least 90% accuracy on the list. For expository text, based on pilot study
data collection, the assessors used their judgment based on the narrative text to start at the
narrative finishing level or one level below. For each text read, the assessor determined
the prior knowledge of the student for the text topic by asking the concept questions
52
52
Pag
e52
2
2 P
age5
22
2
Pag
e52
22
followed by the student reading aloud predetermined narrative and expository passages.
All reading errors and times for reading were recorded to determine accuracy and rate.
The rate was not included in this study as this measure does not identify unique
information about students’ actual accuracy and rate independent of the text read. Upon
completion of oral reading, the student was asked a series of explicit (clearly stated in the
text) and implicit (implied or suggested, but not specifically stated in the text) questions.
The students read passages until the highest instructional reading level was determined
for each student with a minimum of 90% accuracy and 70-85% of the comprehension
questions answered correctly. Leslie and Caldwell (2006) reported inter-rater reliabilities
of .99 for oral reading miscues and .98 for comprehension. For data collection purposes,
the passages used span from QRI levels one to six.
Fluency
Fluency is a student’s ability to read text with sufficient speed, accuracy and
expression to support comprehension. The Dynamic Indicators of Basic Early Literacy
Skills (DIBELS) Oral Reading Fluency (DORF) (Good & Kaminski, 2011) assessment
was used to measure fluency of grade level text. Although the QRI can be used to
measure fluency of instructional level text, there is a specific need to understand student
reading fluency of grade level text with the demands of the ELA Common Core State
Standards (2010). The DIBELS DORF is a standardized, individually administered test
of accuracy and fluency with a set of three passages that are meant to identify children
who may need additional instructional support. It is also used to monitor progress of
students over time with grade level text. Each assessment measures the number of
53
53
Pag
e53
2
2 P
age5
32
2
Pag
e53
22
correct words and accuracy of a students’ reading one minute of grade level text. Student
performance is measured by having students read three passages aloud for one minute
each followed by the student retelling what was read. Words omitted, substituted, and
hesitations of more than three seconds are scored as errors, while self-corrections are
scored as accurate. The number of correct words per minute from the passage is the oral
reading fluency rate. The DORF assessment is a required assessment for students in
grade K-3 as a part of the state assessment framework; however, two of the schools
currently administer the assessment to students in grades K-5. School two does not
administer the assessment to all students K-5. The researchers administered this
assessment to the students.
Good and Kaminski (2002) reported test-retest reliabilities for elementary
students ranging from .92 to .97 with alternate-form reliability of different reading
passages drawn from the same level ranging from .89 to .94. Criterion related validity
has been found in the range of .66-.77 for word correct and .54-.68 for accuracy. This
assessment was used to measure phonics (word recognition) and fluency (accuracy) for
the participants.
Vocabulary
Vocabulary has been shown to be a predictor of reading comprehension. In order
to acquire information on vocabulary knowledge independent of decoding ability, the The
Peabody Picture Vocabulary Test-Revised (PPVT) (Dunn & Dunn, 2007) was
administered to each student. Buly and Valencia (2002) and Dennis (2013) used this
assessment as it measures receptive vocabulary knowledge not dependent upon the
54
54
Pag
e54
2
2 P
age5
42
2
Pag
e54
22
students’ decoding abilities. The assessor started the assessment with pictures that
represent the students’ suggested knowledge based on his/her age. The assessor said a
term while the student looked at four pictures and identified the picture that matched the
term. The assessor continued to present pictures until the student missed eight items in a
section. The PPVT provides a norm scaled score and grade equivalent with a test retest
reliability of .77 for standard scores. This assessment was used to measure vocabulary
for the participants.
Reading Proficiency
The North Carolina English Language Arts (ELA) READY End-of-Grade
Assessments (EOG) is a curriculum-based achievement test administered in grades 3–8.
Third grade students participate in a Beginning of Grade (BOG) assessment. Only fourth
and fifth grade students have EOG scores to determine proficiency for the study, so BOG
scores for third graders were used to determine their proficiency. The ELA/Reading
assessments are aligned to the Common Core State Standards (2010). The ELA
assessment is administered in a paper-and-pencil format with an initial time allotted of
180 minutes with up to an additional 60 minutes testing time if needed. The reading
selections are comprised of narrative and expository selections based on the Common
Core State Standards. Knowledge of vocabulary is assessed indirectly through
application and understanding of terms within the context of the selection and questions.
The EOG assessments of ELA/Reading at grades 3–5 contain 52 total test items, while
the BOG contains 44 items. Access to students BOG/EOG scores for this assessment are
55
55
Pag
e55
2
2 P
age5
52
2
Pag
e55
22
not available to researchers per district policies. Because of this, the schools identified
candidates for participation with BOG/EOG scores of a level 1 or 2.
Motivation
Motivation is complex and domain specific (Wigfield, Guthrie, Tonks, &
Perencevich, 2004), which implies that it can change based on content area and thus will
need to be assessed with instruments that are specific to assessing motivation in that
content area. The MRQ is a domain-specific instrument used to assess the
multidimensionality of reading motivation. In comparison to other motivation
instruments, the MRQ addresses multiple dimensions of motivation within the area of
reading. It has been used by previous researchers to assess students’ motivation and with
its ability to measure multiple motivations for reading, (Guthrie, et al., 2007; Unrau &
Schlackman, 2006; Wang & Guthire, 2004; & Wigfield & Baker 2009) it was the best
option for my dissertation.
The MRQ is a student self-rated assessment of the extent to which each student is
motivated to read (Wigfield & Guthrie, 1997). It contains 53 items that each student
completes independently within a group of approximately 10-15 students (see Appendix
A for MRQ items). The response format ranges from 1= “very different from me” to 4=
“a lot like me.” Scores are computed for each construct by averaging across each of the
respective 11 constructs: Reading Efficacy, Reading Challenge, Reading Curiosity,
Reading Involvement, Importance of Reading, Reading Work Avoidance, Competition in
Reading, Recognition for Reading, Reading for Grades, Social Reasons for Reading, and
Compliance. The assessment began with two practice items and then students completed
56
Pag
e56
2
2 P
age5
62
2
Pag
e56
22
56
Table 2
Instruments for Data Collection
Area to be Assessed Assessment Scores Used
Phonemic Awareness Elision Test: Comprehensive Test of Phonological Processing Raw Score
Phonics DIBELS (Dynamic Indicators of Basic Early Literacy Skills) DORF
Assessment Accuracy Scores
Standard Score
Vocabulary Peabody Picture Vocabulary Test-Revised Standard Score
Grade Equivalent
Fluency DIBELS (Dynamic Indicators of Basic Early Literacy Skills) DORF Standard Score
Comprehension Qualitative Reading Inventory-4. (Narrative and Expository) Grade Equivalent
Scores
Reading Proficiency NC Ready EOG Scale Scores
Proficiency Levels
Motivation Motivation for Reading Questionnaire Domain Average
Scores
57
57
Pag
e57
2
2 P
age5
72
2
Pag
e57
22
the remainder of the questionnaire on their own while the assessor read each item aloud.
Students finished the assessment in one 20-25 minute session. Reliabilities for the
instrument have ranged from .52 to .81. This assessment was used to measure the
motivational values of this sample.
Data Analysis
The study used the following quantitative methods for generating the profiles of
the suggested sample: descriptive statistics (means, standard deviations), factor analysis
(exploratory and confirmatory), and cluster analysis (hierarchical and non-hierarchical).
The study followed the following steps to generate answers to each of the research
questions: 1) identification of descriptive statistics, identification of reading factors
(construct), 2) identification of reading clusters, 3) validation/identification of MRQ
instrument domains, including adjustments to identify best model and domains for the
MRQ, 4) identification of motivation clusters, and 5) identification of reading and
motivation clusters (profiles).
Reading Profiles
First, descriptive statistics were run to provide summaries about the sample and
the measures used. This included cross tabulations to identify the gender and ethnicity,
and grade statistics of the sample (gender, race, grade level). Next, I identified the means
and standard deviations for each grade level, as well as the entire sample, to determine if
the sample fell below grade level expectations for each reading measure. Last, I used
these descriptives to make general conclusions about the sample prior to subsequent
analyses.
58
58
Pag
e58
2
2 P
age5
82
2
Pag
e58
22
Next, an exploratory factor analysis was run for the reading data (Word list,
CTOPP, PPVT standardized scores, QRI Narrative and Expository, and DIBELS correct
words per minute and accuracy scores) using SAS software (SAS Institute, 2011).
Exploratory factor analysis is a data reduction technique that explains correlations
through unknown, unobserved (summary variables) factors (Timm, 2002). This process
identified factors, or unobserved variables, that produced the measured variables.
Although existing research (including my pilot study) have determined factors, the
exploratory factor analysis was used to determine if there are consistent underlying
factors that are represented within this sample from previous work.
Several steps were completed to accurately determine the factors representing the
data. The first step was to determine the number of factors to retain. When doing this
step, multiple methods were used to ensure the best selection (Henson & Roberts, 2006),
including the Kaiser criterion, % of variance explained, and SCREE test. The Kaiser
method determined the number of factors by retaining all factors with eigenvalues above
1 (Costello & Osborne, 2005). For the percentage of variance explained method, the
researcher examined the variance explained by each variable and identified an acceptable
percentage. For this study, the acceptable level was percentage explaining above 65% of
the variance of the data. In the SCREE test method, the researcher visually examined the
scree plots graphical representation of the eigenvalues and identified the area where there
was an elbow, or sudden drop (Cattell, 1966). After determining the factors that
represented the data, the communalities were examined. This process examined the
variance of each item that is accounted for by the factors. The higher the value of the
59
59
Pag
e59
2
2 P
age5
92
2
Pag
e59
22
communality for a variable, the more of its variance was explained by the factors, with
the goal to explain as much variance as possible. Communalities are the sum of the
squared factor loadings for all the factors. When examining the communalities
extractions, those greater than .60 are acceptable, as this criterion signifies that more than
half the variance for that item is explained by the factor. If there are variables with less
than 60% of the variance explained, these variables were deleted from the analysis.
Next, I examined the factor loadings using a Varimax rotation because it
maximized the variance of the loadings for each factor and provided the most
interpretable matrix (Kaiser, 1958). After determining if there were clear and
interpretable loadings for each variable on a factor, I named the underlying reading
constructs that represented this sample. Then, I saved the factors as variables to be used
to understand the profiles of this group of students within the subsequent cluster analysis.
The factor scores representing the reading measures were then used within a
cluster analysis to generate the reading profiles of non-proficient students. Cluster
analysis is a technique that partitions a set of observations/variables into a distinct
number of unknown groups in a manner that all observations within a group are similar,
while observations in different groups are dissimilar (Timm, 2002). It has been
suggested that when true numbers of clusters are unknown a priori, like in the current
study, to use a combination of hierarchical and non-hierarchical clustering techniques to
formulate groups (Sharma, 1996).
Multiple hierarchical methods were used to identify the number of clusters and
cluster membership (profiles). First, I completed the hierarchical cluster analysis using
60
60
Pag
e60
2
2 P
age6
02
2
Pag
e60
22
average, complete, single, and ward linkage to group the variables. In doing cluster
analysis, it is vital to investigate the groups using multiple grouping measures. Then,
dendograms (icicle plots) were examined to determine the number of profiles present
with each grouping method. This process involved examining clusters joined to visually
decide where the largest distance between groups occurs and to stop joining groups at
that point (Rencher & Christensen, 2012). Next, the cluster history was examined to
identify the best cluster solution which had the largest distance between clusters while
keeping the smallest root-mean-square standard deviation (RMSSTD) and Semi-partial
R-Squared values and the highest R-squared values (Rencher & Christensen, 2012),
measures used to determine the homogeneity within the clusters. This procedure required
the researcher to examine each line of the cluster history table, comparing values of the
line to those before and after to determine where the largest “jumps” occur and use the
cluster solution above this. From these two procedures, the amount of clusters were
chosen that best represented the data. After the number of clusters were determined from
the Hierarchical Cluster method, the cluster memberships were refined using non-
hierarchical k-means clustering (Timm, 2002). The means from the hierarchical clusters
were generated in the previous method and were used as initial seeds to refine group
membership for the clusters in the non-hierarchical methods. From this process, the
group membership was defined for each cluster. I then used this information to generate
descriptive statistics of each profile to analyze the patterns present for these non-
proficient students.
61
61
Pag
e61
2
2 P
age6
12
2
Pag
e61
22
Motivation Profiles
The MRQ has theoretical support for its structure (Baker & Wigfield, 1999;
Wigfield & Guthrie, 1997). Although the factor structure has a theoretical basis
established by the authors, my desired sample of students differed from the initial sample.
Because of this, prior to using this instrument to understand the motivation for reading of
these students, it was vital to confirm/identify the factor structure represented in this
sample (Yuan, 2005). First, I used LISREL 9.2 software to determine if the structure of
the MRQ is a fit for my data set. This step was important because the researcher needs to
be confident of the applicability of these motivation items to this sample. Goodness of fit
indices/statistics were used in the CFA to determine how well the MRQ’s a priori model
fits these data (McDonald & Ho, 2002). In using CFA, I determined if the model was a
good, marginal, or Poor fit for the data set (Kline, 2010) and if a Poor or marginal fit,
made modifications to make this a good fit. There are many indices produced in
structural equation modeling, but only certain indices were interpreted. There are no
required indices to be included in decision making for structural models; however, it is
necessary to report a variety of indices to capture different parts of the model fit and
make reasonable determinations (Crowley & Fan, 1997). Kline (2010) suggests inclusion
of the Chi-square indices, root mean square error of approximation (RSMEA), the
standardized root mean square residual (SRMR), and comparative fit index (CFI) to make
decisions about structural model fit. To analyze structural models for this data set, those
indices were used for interpretation of model fit. The Chi-square value is a traditionally
utilized fit index to evaluate a model’s fit. With this index, a good fit generated an
62
62
Pag
e62
2
2 P
age6
22
2
Pag
e62
22
insignificant finding < .05 (Kline, 2010). The root mean square error of approximation
(RMSEA) explained how well the model with estimates fit the covariance matrix of the
sample. For this index, good fit values ranged from .05-.08 (Steiger, 2007). The
RMSEA also produced lower (LB) and upper bound (UB) limits. The upper bound limit
was included and this should be preferably less than .08 for a good fit. The standardized
root mean square residual (SRMR) was used since the questionnaire assesses consistent
levels i.e. 1-4 (Kline, 2010). Good fit values for SRMR were less than 0.08 (Hu &
Bentler, 1999). The comparative fit index (CFI) assessed the model fit. Values larger
than .90 are acceptable with a requirement of larger than .95 for a good fit (Hu & Bentler,
1999).
After identifying items and factors that are represented in my sample, I named the
factors. Next, I generated and discussed descriptive statistics representing the means of
individual items and the means of the factors. These descriptives were used to make
general conclusions about this sample of non-proficient readers’ motivation for reading
prior to subsequent analyses. I then completed the same cluster analysis process with the
motivation average scores for each factor. This allowed me to generate profiles that can
be compared to the existing work of Baker and Wigfield (1999).
Reading and Motivation Profiles
Last, I used the reading factor scores and the motivation factor scores to generate
profiles that represented the reading and motivation patterns of these non-proficient
students. These profiles were generated with the same procedures as done for the reading
and motivation profiles in isolation. These profiles were generated as interpretations of
63
63
Pag
e63
2
2 P
age6
32
2
Pag
e63
22
the reading profiles as reading and motivational factors clustered together. These new
profiles were analyzed with mean scores and other descriptive statistics that represent the
factors to determine the patterns that exist with these students from multiple analyses.
An analysis was also completed to compare differences between the single and combined
profiles.
64
64
Pag
e64
2
2 P
age6
42
2
Pag
e64
22
CHAPTER IV
RESULTS
This study focused on understanding the reading and motivation profiles for
students, who were not successful on their state mandated reading assessment. This
included identifying underlying reading and motivational constructs and the
multidimensionality of these constructs. The data collected measured students’ skills in
reading by assessing the five major areas of reading: phonemic awareness, phonics (word
recognition), fluency, vocabulary and comprehension; and motivations for reading:
including efficacy, challenge, avoidance, curiosity, involvement, importance, recognition,
grades, competition, social, and compliance. The instruments used to collect the data
were discussed in the previous chapter.
I first discuss the reading descriptive data by analyzing each school, by grade
level, and the entire sample. This is followed by the results of the exploratory factor
analysis and cluster analysis. Then, I discuss the motivation descriptive data. This is
followed by results from a confirmatory factor analysis and cluster analysis of the
motivation data. Last, I present the results from the combined cluster analysis,
representing data from both the motivation and reading clusters. I have structured the
results analysis this way because previous studies have not examined profiles from a
65
65
Pag
e65
2
2 P
age6
52
2
Pag
e65
22
combination of reading motivation, but in isolation. This organization of the results
presents initial results that can be evaluated to determine if the existing profiles and
patterns are consistent with my sample.
Reading Trends
In this section, I present descriptive information by grade and school level for
each of the reading variables assessed. This data was generated by grouping data based
on grade levels and schools to identify mean scores that could be used to examine
patterns present. A multivariate ANOVA (MANOVA) was conducted with the reading
variables to determine if significant differences existed between the school level data.
Table 3 presents descriptive information (means, standard deviation, and skew) of
student performance across the schools for the seven reading variables. From the seven
variables, all appear to be within a normal range except DIBELS accuracy scores which
are negatively skewed. With this minor lack of normality which can be attributed to the
nature of a population inclusive of only non-proficient students, the data is acceptable to
future analyses.
When examining Table 3, four general patterns appeared. First, as expected,
given the non-proficient status of the students, performances overall were below grade
level expectations. For each variable, there were grade level expectations that were not
met within the sample for grades 3, 4, and 5 respectively, DIBELS expected accuracy
96%, 97%, & 98%; DIBELS correct words per minute 86, 103, and 112; QRI scores for
comprehension and the word list should be greater than or equal to the grade level.
Second, schools had different patterns of high and low scores in comparison to each
66
66
Pag
e66
2
2 P
age6
62
2
Pag
e66
22
other. Of the seven variables assessed, School 3 had four variables with the highest
scores (QRI-Word list, DIBELS WPM, and QRI-narrative and expository). The other two
schools had fewer variables with the highest scores. School 1 had two with the highest
scores (CTOPP and PPVT) and School 2 had one (DIBELS Accuracy). Third, with
exception of CTOPP and DIBELS accuracy, School 2 had the smallest standard
deviations of most of the variables assessed. All though this school had the most students
in the sample, it provided the smallest spread of data. Last, each school had consistent
differences between QRI expository and narrative text comprehension. The expository
comprehension levels were approximately one grade level lower than their narrative
scores. This difference suggests these students were not comprehending expository texts
at the same grade level as they were narrative texts.
67
67
Pag
e67
2
2 P
age6
72
2
Pag
e67
22
Table 3
Descriptive Statistics for the Reading Data by School
Variable School
1
n=64
School
2
n=77
School
3
n=46
Average of
Scores
n=187
Skew
Phonemic Awareness
CTOPP* 14.77
5.06
9.94
4.35
11.61
4.14
12.50
4.99
.01
Phonics/Word ID
DIBELS ACC *** 93.86
11.55
96.09
6.08
95.80
5.16
95.07
8.58
-5.18
QRI Word List **** 3.04
1.65
3.33
1.41
3.75
1.52
3.35
1.57
.13
Vocabulary
PPVT GE**** 3.08
1.67
2.74
1.58
3.05
1.61
2.99
1.63
1.10
Fluency
DIBELS 82.49
37.34
86.54
29.71
95.08
37.27
87.80
35.83
.02
Comprehension
QRI E **** 1.83
1.06
1.59
.79
1.98
.98
1.82
.97
1.25
QRI N **** 2.70
1.16
2.72
.98
2.88
1.19
2.77
1.13
.41
Note: E=Expository Text, N=Narrative Text, *Raw Score, ***Percent Correct,
****Grade Equivalent, number in italics is the standard deviation
1. Phonemic Awareness: The raw scores for CTOPP range from 1-20 with
average scores falling between 8 and 12.
2. Phonics/Word Identification: Percent correct, DIBELS expected accuracy
96%, 97%, and 98% for grades 3, 4 & 5 respectively. QRI word list
expectation is to be able to successfully read grade level word list.
3. Vocabulary: PPVT GE expectation is to identify vocabulary at grade level.
4. Fluency Scores: Correct Words read per Minute Expected scores for
DIBELS are 86, 103, and 112 for grades 3, 4 & 5 respectively.
5. Comprehension: For the QRI all scores representative of the highest
instructional level text read by grade level equivalents for the passages.
68
68
Pag
e68
2
2 P
age6
82
2
Pag
e68
22
In addition to the patterns and differences observed in the reading data, a
MANOVA was completed to determine if there were significant differences for the seven
variables between the three schools. Table 4 provides the results of the MANOVA test.
There was evidence of a significant difference between schools for at least one of the
reading variables F(14, 346) = 5.26, p < .0001. As a result of the MANOVA results, the
univariate ANOVA tests were completed to determine which variables were significantly
different between schools. When examining the univariate tests, significant differences
were noted for the word list F (2, 184) = 3.68, p < .0271 and the CTOPP F (2, 184) =
17.76, p < .0001. A post hoc test was completed using the Tukey HSD test to determine
which means were different between the schools for the CTOPP and Word List variables
(see Table 5). For the word list variable, a significant difference was found between
School 1 and School 3 (p=.02). There were no significant differences between School 1
and School 2 (p=.58), nor between School 2 and School 3 (p=.34). For the CTOPP
variable, significant differences were found between School 1 and School 2 (p=.00), and
School 1 and School 3 (p=.00). No significant difference was found between School 2
and School 3 (p=.15). These results showed that School 1 is higher than School 2 and 3
for the CTOPP; while for the Word list, School 1 is lower than School 3.
69
69
Pag
e69
2
2 P
age6
92
2
Pag
e69
22
Table 4
MANOVA Test Results for Reading Differences Between Schools
Statistic Value F Value Num DF Den DF
Wilks' Lambda .69 5.26* 14 356
Pillai's Trace .33 4.96* 14 358
Hotelling-Lawley Trace .44 5.57* 14 281
Roy's Greatest Root .40 10.19* 7 179
Note: *= p<.05. F. Dependent Variable=School, Independent Variable(s) = Reading
Assessments.
Table 5
Post Hoc Test Comparison Between Schools
Dependent Variable School School Mean
Difference
Std. Error
Word List
1 2 -.29 .29
1 3 -.71* .26
2 3 -.42 .30
CTOPP
1 2 4.83* .86
1 3 3.16* .78
2 3 -1.68 .89
Note: TUKEY HSD used for comparisons. *=p<.05.
70
70
Pag
e70
2
2 P
age7
02
2
Pag
e70
22
Table 6
Descriptive Statistics for the Reading Data by Grade Level
Variable 3rd
n=83
4th
n=54
5th
n=50
Average of
Scores
n=187
Phonemic Awareness
CTOPP* 11.76
4.95
12.26
5.07
13.98
4.74
12.50
4.99
Phonics/Word ID
DIBELS ACC *** 94.25
7.61
94.17
12.21
97.40
3.66
95.07
8.58
QRI Word List **** 2.89
1.42
3.28
1.55
4.20
1.53
3.35
1.57
Vocabulary
PPVT GE**** 2.54
1.28
2.82
1.66
3.90
1.76
2.99
1.63
Fluency
DIBELS 76.84
34.50
86.93
33.94
106.92
32.54
87.80
35.83
Comprehension
QRI E **** 1.45
.69
1.89
.77
2.38
1.26
1.82
.97
QRI N **** 2.31
.91
2.70
.94
3.58
1.20
2.77
1.13
Note: Scores combined for grade levels represent all schools. E=Expository Text,
N=Narrative Text, *Raw Score, ***Percent Correct, ****Grade Equivalent, number in
italics is the standard deviation.
1. Phonemic Awareness: The raw scores for CTOPP range from 1-20 with
average scores falling between 8 and 12.
2. Phonics/Word Identification: Percent correct, DIBELS expected accuracy
96%, 97%, and 98% for grades 3, 4 & 5 respectively. QRI word list
expectation is to be able to successfully read grade level word list.
3. Vocabulary: PPVT GE expectation is to identify vocabulary at grade level.
4. Fluency Scores: Correct Words read per Minute Expected scores for DIBELS
are 86, 103, and 112 for grades 3, 4 & 5 respectively.
5. Comprehension: For the QRI all scores representative of the highest
instructional level text read by grade level equivalents for the passages.
71
71
Pag
e71
2
2 P
age7
12
2
Pag
e71
22
Table 6 presents descriptive information (means and standard deviations) of
student performance across grade levels for the seven reading variables. With minimum
differences between school level data, and the uneven numbers of students, grade level
descriptive statistics were combined across schools and then evaluated. When examining
this table, five general patterns were identified. First, as observed in the school
comparisons, performances for each grade level were below grade level expectations.
For each variable, there were grade level expectations that were not met within the
sample or grades 3, 4, and 5 respectively DIBELS expected accuracy 96%, 97%, & 98%;
DIBELS correct words per minute 86, 103, & 112; QRI scores for comprehension and the
word list should be greater than or equal to the grade level. Second, except for DIBELS
accuracy scores, as expected, students’ performance for the reading scores increased by
grade level. Third, students’ ability to manipulate sounds of words (CTOPP), read words
in isolation (Dibels ACC and QRI Word Lists), and identify single word meanings
(PPVT) were closer to grade level than were their ability to read quickly (DIBELS) and
comprehend texts, particularly for expository passages. Fourth, students’ ability to
recognize and decode words were at a higher level than their comprehension levels. The
word list scores were all less than a grade level lower than expected based on grade level
placements. Last, when examining the comprehension variables, the spread (standard
deviation) grew as students increased in grade levels. This shows that the comprehension
of non-proficient students widens as students continue to higher grade levels, and this
area needs a greater emphasis than word recognition for these students. This finding
signifies that all non-proficient students were not the same and variances grow as
72
72
Pag
e72
2
2 P
age7
22
2
Pag
e72
22
students proceed to higher grade levels. A MANOVA was not completed for
comparisons between grade levels because the expectation was for differences in
performances for each of the variables between grade levels.
In summary, the descriptive data presented in this section highlighted many of the
patterns present for the sample. The students were consistently below grade level
expectations regardless of the school. Only minimal significant differences were found
between mean scores for each school. These initial descriptive patterns provided the
introduction to later findings by identifying reading trends represented by these 3rd, 4th
and 5th grade non-proficient students. The initial differences suggested patterns of
performance for students, but with many individual characteristics to consider. The
larger standard deviations provided an initial support for the idea that non-proficient
readers were not identical in their needs. For the rest of this chapter, with the exception
of the motivation item level descriptive summaries, the schools were combined for
analysis. With no major differences and consistent trends in their performance, the data
was suitable for combining. A factor analysis determined if these multiple reading
characteristics could be reduced to fewer constructs, which were easier to interpret and
group. Then, these constructs were used to group students via the clustering process into
groups with similar patterns of performance.
Reading Factors
To better understand students’ strengths and weaknesses, factors were generated
to reduce the data into a smaller subset of interpretable values. This process identified
factors, or unobserved variables, that explained a large percentage of the measured
73
73
Pag
e73
2
2 P
age7
32
2
Pag
e73
22
variables. An exploratory factor analysis was performed using SAS software with the 11
variables to identify these unobserved factors.
The first step of the exploratory factor analysis determined the number of factors
to retain. When doing this step, multiple criteria were examined to ensure the best
selection (Henson & Roberts, 2006). Three methods determined the factors--Kaiser
criterion, SCREE test (Cattell, 1966), and percent explained. The Kaiser method
determined the number of factors by retaining all factors with eigenvalues above 1
(Costello & Osborne, 2005). In the SCREE test method (see figure 3) the researcher
visually examined the scree plots graphical representation of the eigenvalues and
identified the area where there is an elbow, or sudden drop (Cattell, 1966). The
percentage method explained the number of factors to retain that represent a high
percentage of the variance. Using these three methods, a two-factor solution was
determined. The scree plot displayed a drop after two factors. While there is only one
eigenvalue above 1.0, the second is close with .93 and the two-factor solution explained
72% of the total variance which is acceptable (see Table 7).
74
74
Pag
e74
2
2 P
age7
42
2
Pag
e74
22
Figure 3. Scree Plot of the Reading Factors
Table 7
Total Variance of Reading Factors
Component Initial Eigenvalues
Total % of Variance Cumulative %
1 3.40 56.73 56.73
2 .93 15.55 72.29
3 .72 11.96 84.25
4 .48 7.95 92.20
5 .26 4.36 96.56
6 .21 3.44 100.00
Next, the communalities were examined. This process examined the variance of
each item that was accounted for by the factors. The higher the value of the communality
for a variable, the more of its variance is explained by the factors, with the goal to explain
as much variance as possible. When examining the communalities extractions, those
75
75
Pag
e75
2
2 P
age7
52
2
Pag
e75
22
greater than .60 were acceptable, as this criterion signifies that more than half the
variance for that item was explained by the factors. Variables with less than half of their
variance explained by the factors were not accurately measured by the factors. With the
initial two-factor model, all loadings were greater than .6, with the exception the CTOPP
(Table 8). Similar to the work of Buly and Valencia (2002, 2004), the CTOPP did not
have more than half of the variance of the variable explained by the factors and was
removed from subsequent analyses. When running the factor analysis again without the
CTOPP scores, all extractions were greater than .60 (see Table 8). This supported the
two-factor solution for this data set.
Table 8
Reading Factor Communalities
Initial Extraction Revised Extraction
Word List .78 .78
CTOPP .33 Removed
PPVT Standard .73 .68
QRI- Narrative .64 .69
QRI- Expository .66 .74
DIBELS WPM .79 .79
DIBELS Accuracy .63 .66
Note: BOLD is below 50% of the variance explained by the factor structure.
76
76
Pag
e76
2
2 P
age7
62
2
Pag
e76
22
The next step involved examining the factor matrix to determine if there were
unequivocal loadings for each variable on a factor. This step ensured that the two factors
are actually interpretable with representation of the variables. For this step, a Varimax
rotation (see Table 9) was used to determine which variables loaded onto which factor,
because it maximized the variance of the loadings for each factor and provided the most
interpretable matrix (Kaiser, 1958). The Varimax rotation produced all acceptable
loadings, with the lowest being .60. Next, with interpretable factors, the factors were
appropriately named, word recognition and meaning, similar to that of Buly and Valencia
(2002, 2004) and Dennis (2013).
Table 9
Reading Variables Factor Loadings
Component
Word Identification Meaning
Word List .81 .35
PPVT Standard .00 .82
QRI- Narrative .57 .60
QRI- Expository .42 .75
DIBELS WPM .85 .26
DIBELS Accuracy .81 .02
Note: Bold Numbers: Heaviest loadings for each factor. Rotation Method:
Varimax with Kaiser Normalization.
77
77
Pag
e77
2
2 P
age7
72
2
Pag
e77
22
Factor 1: Word Identification
Factor one was labeled as word identification, as each of the measures that loaded
onto this factor were related to word recognition abilities, including fluency and rate.
Each loading was greater than .80 which indicates a higher percentage of this variability
explained by the factor. These loadings included DIBELS accuracy scores, CWPM
scores, and QRI word list scores. This factor was attributed to 56.73% of the total
variance. The items empirically and theoretically fit together on this factor because
accuracy and rate are often associated with each other in reading assessments.
Factor 2: Meaning
Factor two was labeled as meaning because each of its measures were inclusive of
students’ ability to make meaning from knowledge of vocabulary and making meaning
from texts read. The QRI measures of comprehension for expository and narrative text as
well as the PPVT loaded here. Meaning was attributed to 15.55% of the total variance.
Factor loadings for this factor were all above .6, thereby supporting this factor as
representing a large amount of variability of each loading. Although the PPVT does not
require students to comprehend a text, it loaded highly onto this factor. This assessment
is a test of receptive vocabulary and measures students’ ability to understand and identify
items.
The results of the exploratory factor analysis generated two factors, word
recognition and meaning, which represented the underlying reading constructs for this
group of students. However, factor analysis does not provide information on patterns of
78
78
Pag
e78
2
2 P
age7
82
2
Pag
e78
22
performance for understanding the similarities and differences present within this group.
The next analysis addressed this question.
Reading Profiles
Cluster analysis was performed on the reading factors, word identification and
meaning, to generate profiles/clusters that represent non-proficient readers. The cluster
analysis is a statistical technique that divided students into similar groups (Timm, 2002).
It allowed the researcher to interpret the sample by identifying patterns amongst similar
groups of students. It has been suggested that when true numbers of clusters are
unknown a priori, like in the current study, the researcher should use a combination of
hierarchical and non-hierarchical clustering techniques to formulate groups (Sharma,
1996). Both clustering techniques were used in this study.
The SAS program was used to perform hierarchical and non-hierarchical cluster
analysis to identify the groups of students with similar patterns. The non-hierarchical
cluster analysis was completed to determine the number of clusters, without a specific
focus on the group membership. Average, complete, single, centroid, and Ward’s linkage
methods were used within the analysis to group the data. These methods were used
because they provided different ways to group the variables by the nearest neighbor or
chaining method (single), furthest neighbor (complete), average distance between clusters
(average), all possible cluster pairs combined and summed (Ward’s), and the mean value
for each variable is calculated and used for clustering (centroid). The use of these
methods also helped to establish consistency when determining the number of clusters
present.
79
79
Pag
e79
2
2 P
age7
92
2
Pag
e79
22
First, dendograms and cluster group history were examined using average,
complete, single, centroid, and Ward’s linkage methods to determine the number of
clusters. The dendograms were examined to visually determine the number of groups by
identifying the groups with the smallest distance between them. This process involved
examining clusters joined to visually determine where the largest distance between
clusters occurs. This method provided less consistent results than desired. Single linkage
methods were uninterpretable visually due to the large sample size. The average and
centroid methods visually displayed five groups, while the ward and complete methods
displayed four groups.
Next, the cluster histories were examined to identify a solution which would have
the largest distance between clusters while keeping the smallest RMSSTD and Semi-
partial R-Squared values and the highest R-squared values. This procedure required the
researcher to examine each line of the cluster history table (see Table 10), comparing the
values to those of the line before and after to determine where the largest “jumps”
occured. Then, the cluster solution above this one was selected. I reported the average
method, as this was the method used to perform the hierarchical step of the clustering
procedure. For this method, the largest jumps occurred between lines three and four for
the RMSSD, Semipartial R-squared, and R-Square values, while allowing for the smallest
RMSSD and Semipartial R-squared values and the largest R-Square values. As a result, a
four-cluster solution was selected as the best representation for the data.
With a number of clusters selected, I completed hierarchical clustering of the data.
This step refined group memberships to determine the exact number of students in each
80
80
Pag
e80
2
2 P
age8
02
2
Pag
e80
22
group. The initial seeds (mean scores) from the non-hierarchical stage were used as seeds
to sort students into groups based on those means. This process ensured students in each
group were more similar to students in the same group and not as similar to other groups.
Table 10
Reading Cluster History
Number
of
Clusters
Clusters Joined n New
Cluster
RMS Std
Dev
Semipartial
R-Square
R-
Square
Maximum
Distance
10 CL22 CL23 21 .41 .01 .83 1.22
9 CL15 CL14 45 .46 .02 .82 1.25
8 CL17 CL12 86 .55 .05 .77 1.38
7 CL10 CL16 31 .56 .03 .74 1.42
6 CL9 CL8 131 .64 .10 .64 1.52
5 CL7 CL41 33 .61 .01 .63 1.73
4 CL6 OB69 132 .65 .01 .62 2.19
3 CL4 CL11 153 .82 .23 .39 2.42
2 CL5 CL3 186 .96 .32 .08 2.52
1 CL2 OB38 187 1.00 .08 0.00 5.48
Note: Complete Clustering Method Used. Only possible clusters less than 10 were
included to save space. Shaded line identifies greatest “jumps” and the best cluster.
81
81
Pag
e81
2
2 P
age8
12
2
Pag
e81
22
Table 11
Reading Clusters Summary Data
Cluster
n Maximum
Distance
from Seed to
Observation
Distance
Between
Cluster
Centroids
Cluster Means
WordID Meaning
1 49 2.21 1.44 .45 1.25
2 35 4.03 2.02 -1.58 .06
3 52 1.75 .97 .53 -.18
4 51 1.50 .97 .11 -1.06
The revised groups contained students who were distinct from the other clusters,
as shown by the maximum distance from seed to observation, and the distance between
cluster centroid columns of the cluster summary table above (see Table 11). Several
patterns emerged from the clusters, as noted in the cluster summary, with most of the
clusters having patterns of high and low scores with word identification and meaning.
While examining these patterns, they must be interpreted within the context of these
students still performing below grade level expectations.
Cluster one featured 49 students who appeared as most successful with both word
identification and meaning as each of these scores are above the median. These students
had the highest meaning scores of the sample. While their word identification scores
were not the highest, they were still high based on a within group comparison. The
smallest cluster membership was for the students in cluster two. These 35 students had a
combination of low word identification and higher meaning scores. These students had
82
82
Pag
e82
2
2 P
age8
22
2
Pag
e82
22
the lowest word identification scores for the sample and meaning scores slightly above
the median.
More than half of the sample (n=103) were grouped in clusters 3 and 4. These
two clusters have combinations of students, who had some of the lowest scores for the
sample in either word identification or meaning, with some scores above the median and
some below the median. Cluster 3 students had the highest word identification scores for
the sample, with meaning scores that just below the median. Cluster 4 students had the
lowest meaning scores for the sample with word identification scores just above the
median.
Motivation for Reading Trends
The Motivation for Reading Questionnaire (MRQ) was used to assess students’
motivation for reading within 11 predetermined dimensions noted by the authors. Table
12 highlights the descriptive statistics for each motivation question including means and
standard deviations. Similar to the reading descriptive data, several patterns emerged
from the motivation data. First, the data does not present motivation overall as high or
low for students across all areas. The students’ responses showed scores ranging from
2.13 to 3.65 across the motivation questions, with a possible score range from 1 to 4. The
question with the lowest score was 11, I visit the library often with my family. The
highest scores, 3.65, were associated with two questions: 17, I know that I will do well in
reading next year and 27, it is very important to me to be a good reader. Second, with a
format ranging from 1= “very different from me” to 4= “a lot like me,” the average range
would be considered to be scores between 2.0 and 3.0. This data set had 27 items with
83
83
Pag
e83
2
2 P
age8
32
2
Pag
e83
22
scores above 3.0 and 26 items with scores between 2.0 and 3.0. These non-proficient
students had more than half of the items with high scores. While it may seem appropriate
to say that the students have a high motivation for reading, this inference cannot be made
as the students agree with statements and some of the motivation domains that are
represented by items that can be considered negative, such as #11, I visit the library often
with my family. This question examined motivation but it does not mean this is
something the student does not enjoy, but something in which they do not participate.
Analyses were run to determine the suitability of the data for further analyses.
The MRQ was found to be highly reliable (53 items; α = .87). The results of item level
Shapiro-Wilk test for normality statistics revealed that each item assessed with the MRQ
failed to be normally distributed. This was expected because the sample was specific to
non-proficient students and not a range of all students in 3rd, 4th, and 5th grades. Although
the data lacked normality, subsequent analyses were completed with the necessary
precautions.
84
84
Pag
e84
2
2 P
age8
42
2
Pag
e84
22
Table 12
Motivation Item Descriptive Statistics
Item
# Mean
Std.
Deviation
Shapiro-Wilk
Statistic
1 3.06 .93 .82*
2 2.66 1.03 .87*
3 3.64 .70 .57*
4 3.04 .89 .83*
5 2.69 1.16 .83*
6 2.73 1.01 .87*
7 3.65 .71 .55*
8 2.91 1.05 .84*
9 3.09 1.12 .75*
10 3.28 .93 .75*
11 2.13 1.15 .81*
12 3.05 1.04 .80*
13 2.63 1.12 .84*
14 2.80 1.12 .83*
15 3.30 .89 .75*
16 3.00 .98 .83*
17 3.65 .66 .59*
18 3.39 .90 .69*
19 3.15 .92 .81*
20 2.83 .99 .85*
21 2.66 1.03 .87*
22 2.99 1.05 .82*
23 2.42 1.29 .78*
24 2.37 1.17 .83*
25 3.56 .82 .59*
26 2.50 1.35 .75*
27 3.49 .80 .67*
28 3.64 .73 .55*
29 2.80 1.08 .87*
30 3.12 1.04 .77*
31 2.53 1.21 .82*
32 2.27 1.22 .80*
33 2.94 1.12 .80*
34 2.44 1.17 .83*
85
85
Pag
e85
2
2 P
age8
52
2
Pag
e85
22
Table 12
Motivation Item Descriptive Statistics
Item
# Mean
Std.
Deviation
Shapiro-Wilk
Statistic
35 2.41 1.16 .84*
36 3.44 .89 .66*
37 2.64 1.20 .82*
38 3.56 .80 .60*
39 2.60 1.13 .85*
40 2.88 1.17 .79*
41 3.09 1.09 .77*
42 2.81 1.19 .80*
43 3.41 0.91 .68*
44 3.53 0.85 .61*
45 2.32 1.16 .83*
46 3.38 0.92 .69*
47 3.36 0.89 .72*
48 2.94 1.05 .82*
49 3.03 1.14 .77*
50 3.48 0.85 .65*
51 3.34 0.86 .74*
52 2.54 1.21 .82*
53 3.45 .92 .63*
Note: df=187. *= p<.05. The information is
collapsed across grades and schools and
represents the complete sample.
Motivation Factors
Motivation Factor Structure
Wigfield and Guthrie (1997) demonstrated that the 53 MRQ items load onto 11
reading motivation dimensions with their initial sample. Prior to using this instrument to
analyze this data set, it was vital to confirm that this factor structure (see Table 13) was
represented in this sample (Yuan, 2005). LISREL 9.2 software (Jöreskog & Sörbom,
86
86
Pag
e86
2
2 P
age8
62
2
Pag
e86
22
1996) was used to complete a confirmatory factor analysis (CFA) of the MRQ data from
the dataset (n=187). This step was important because the researcher needed to be
confident of the applicability of these motivation items to this sample. The distribution
of the data did not fall within a normal distribution range as noted by the results of the
Shapiro-Wilk Test (see Table 12 above); however, maximum likelihood estimation was
used in the confirmatory factor analysis to account for the non-normality within the data.
Goodness of fit statistics were used in CFA to determine how well the a priori
model fit the specific data (McDonald & Ho, 2002). When determining the fit, the
researcher seeks to determine if the model was a good, marginal, or Poor fit for the data
set and to identify the acceptable model for interpretation (Kline, 2005). There are many
indices produced in structural equation modeling, but only certain indices were
interpreted in this study. There are no required indices to be included in decision making
for structural models; however, it is necessary to report a variety of indices to capture
different parts of the model fit and make reasonable determinations (Crowley & Fan,
1997). To analyze the structural models for this data set, the SRMR, RSMEA estimate,
RMSEA upper limit of the confidence interval, and CFI were used. Kline (2005) suggests
inclusion of these indices to make decisions about structural model fit.
87
87
Pag
e87
2
2 P
age8
72
2
Pag
e87
22
Table 13
Initial MRQ Factor Structure by Instrument Authors
Factor Items
Curiosity 10, 14, 19, 25, 29, 4
Involvement 6, 12, 30, 33, 35, 22
Competition 9, 41, 44, 49, 52
Social 11, 26, 42, 45, 48, 31, 39
Efficacy 7, 15, 21
Compliance 36, 46, 51, 23, 34
Importance 17, 27
Challenge 2, 5, 8, 16, 20
Avoidance 13, 24, 32, 40
Grades 3, 38, 50, 53
Recognition 18, 28, 37, 43, 47
Each of the above indices were included because of the unique information about
the model that it presented. The Chi-square value is a traditional fit index that evaluates
the model’s overall fit. With this index, a good fit generates an insignificant result such
as > .05 (Kline, 2005). The root mean square error of approximation (RMSEA) explains
how well the model with estimates fit the covariance matrix of the sample. For this
index, good fit values range from .05-.08 (Steiger, 2007). The RMSEA also produces
lower (LB) and upper bound (UB) limits. The upper bound limit will be included as this
88
88
Pag
e88
2
2 P
age8
82
2
Pag
e88
22
should be preferably less than .08 for a good fit. The standardized root mean square
residual (SRMR) is used since the questionnaire assesses consistent levels such as 1-4
(Kline, 2005). Good fit values for SRMR are less than .08 (Hu & Bentler, 1999). The
comparative fit index assesses the model fit, including smaller sample sizes. Values
larger than .90 are acceptable with a requirement of larger than .95 for a good fit (Hu &
Bentler, 1999). The ideal sample size for a CFA in a structural model is 200-400. This
sample fell just short, so there were multiple interpretations of the model to ensure the
best match between motivational domains and representation of the data. First, the model
was run with all parameters simultaneously to determine the fit. This initial model was
the item loadings noted by the authors of the instrument. The full model produced the
following results: SRMR = .07, RMSEA = .05, RMSEA UB = .06, CFI of 0.90, χ2=
1920.98, df = 1270, and p = 0.00. While the Chi-Square p-value is not greater than .05,
the remaining indices were at acceptable and good levels. When examining the
standardized loadings, there were several items (13, 24, 32, 40) that had insignificant
loadings (p >.05) and errors with variances. Due to the poor fit of this data, model fit, the
errors, and significant loadings needed further examination and the model needed to be
examined with subsets of the data.
Baker and Wigfield (1999) had what they noted as an undesirable sample (n=371)
and decided to complete three separate CFA’s of their data. For their model, they used
subsets of the factors, ranging from two to six factors to measure the fit of the model for
their data. With this data set close to the ideal sample size, I needed to examine the
model with two subsets of the factor structure. The first half of the model (Challenge,
89
89
Pag
e89
2
2 P
age8
92
2
Pag
e89
22
Grades, Curiosity, Involvement, Efficacy, and Avoidance) produced results of: SRMR =
.07, RMSEA = .05, RMSEA UB = .06, CFI of 0.88, χ2= 492.34, df = 335, and p = 0.00.
While the Chi-Square p-value was not greater than .05, the remaining indices were at
acceptable and good levels. Similar to the full model, this model produced several
insignificant standardized loadings. Unlike the full model, this model did not produce
any error negative variances. The second half of the model (Competition, Importance,
Recognition, Compliance, and Social) produced a SRMR = .07, RMSEA = .07, RMSEA
UB = .07, CFI of .88, χ2= 472.81, df = 265, and p = 0.00. For this half, the Chi-Square p-
value is not greater than .05, the CFI value was below the acceptable range, however, the
remaining indices were at good levels. Unlike the full model or the other half of this
model, there were no insignificant standardized loadings nor error negative variances.
With the identified issues in both the full and split models, changes needed to be
made to the factor structure. LISREL software outputs information for modification
changes for a better fit for data; however, the suggested changes were determined to not
be realistic based on the actual survey questions and the domains. The avoidance factor
was responsible for the majority of the issues with the model. The scores were reversed
as suggested by the instrument’s authors, but it continued to produce issues in the models
run. The best modification was to delete the avoidance factor and its associated
questions. This was done because this factor was causing the negative variances and the
insignificant standardized loadings. Using this modification to the structure, the full
model produced these calculations: SRMR = .07, RMSEA = .05, RMSEA UB = .06, CFI
of .91 χ2= 1647.04, df = 1082, and p = 0.00. The modification removed the negative
90
90
Pag
e90
2
2 P
age9
02
2
Pag
e90
22
variances as well as the insignificant loadings and produced a good fit for the data. To
verify the fit of the model, the structural model was evaluated with two subsets as
completed earlier. Both models suggest a good fit for the model with the data.
While completing the confirmatory analysis, some challenges developed and were
overcome. A consistent issue with the Chi-Square test was noted as all of the models
continued to produce significant values when an insignificant value is desired. Although
this was an issue, it did not impact the model’s acceptance as this test assumes
multivariate normality, which was not represented for this data set. Second, the large
number of items for the assessment made the path diagram uninterpretable. With this, it
was not included or interpreted. While determining if the factors and items would be
accepted, the loadings were examined for each item. Each item produced significant
loadings. This finding supports the use of the 10 factors for subsequent analyses. Table
14 identifies the final factor structure with the items that load onto each factor.
91
91
Pag
e91
2
2 P
age9
12
2
Pag
e91
22
Table 14
Revised Final MRQ Factor Structure
Factor Items
Curiosity 10, 14, 19, 25, 29, 4
Involvement 6, 12, 30, 33, 35, 22
Competition 9, 41, 44, 49, 52
Social 11, 26, 42, 45, 48, 31, 39
Efficacy 7, 15, 21
Compliance 36, 46, 51, 23, 34
Importance 17, 27
Challenge 2, 5, 8, 16, 20
Motivation Factors
The revised structural model for the data included ten motivation for reading
factors (domains) including Challenge, Competition, Compliance, Curiosity, Efficacy,
Grades, Recognition, Social, Importance and Involvement. These items were included in
the original 11 theoretical domains by the authors of the instrument. The difference was
that the dimension of avoidance was not clearly identified due to insignificant loadings
and variance errors. The ten domains identified were all labeled according to the original
theoretical model by Wigfield and Guthrie (1997) noted in Chapter 3.
92
92
Pag
e92
2
2 P
age9
22
2
Pag
e92
22
Motivation Factor Descriptive Data
Table 15 presents the descriptive information for the motivation domains of each
school. When examining the school performances, none of the scores were extremely
high or low in comparison to those of the other schools. When examining the individual
domains, there was no school that contained all high or low scores for every motivation
domain. School 3 had the highest scores in six domains (efficacy, challenge (tie),
curiosity, importance, grades, and competition). School 1 had the highest in four
domains (challenge (tie), involvement, recognition, and compliance).
In addition to the patterns and differences observed from the chart with
descriptive statistics, a MANOVA test was completed to determine if there were
significant differences between the school level motivation variables. Table 16 provides
the results of the t-tests. There was no evidence that there was an overall significant
difference between the schools for the motivation variables: F (20, 350) = 1.39, p =0.126.
93
93
Pag
e93
2
2 P
age9
32
2
Pag
e93
22
Table 15
Descriptive Statistics for the Motivation Data by School
Variable School 1
n=64
School 2
n=77
School 3
n=46
Total
n=187
Social 2.53
.61
2.56
.60
2.55
.69
2.55
.62
Challenge 2.88
.73
2.73
.64
2.88
.60
2.82
.67
Involvement 2.89
.48
2.77
.60
2.80
.57
2.82
.56
Compliance 3.03
.52
2.99
.48
3.00
.42
3.00
.48
Competition 3.14
.66
2.90
.63
3.21
.63
3.06
.65
Curiosity 3.12
.57
3.05
.65
3.19
.47
3.11
.58
Efficacy 3.28
.55
3.02
.65
3.41
.58
3.20
.62
Recognition 3.40
.54
3.14
.69
3.38
.49
3.29
.60
Grades 3.58
.55
3.41
.69
3.66
.39
3.53
.54
Importance 3.60
.60
3.43
.69
3.75
.42
3.57
.61
Note: Variables have been sorted with total averages in ascending order. Possible
values can range from 1 to 4. Standard Deviations are italicized.
94
94
Pag
e94
2
2 P
age9
42
2
Pag
e94
22
Table 16
MANOVA Test Results for Motivation Differences Between Schools
Statistic Value F Value Num DF Den DF
Wilks' Lambda 0.86 1.39 20 350
Pillai's Trace 0.15 1.37 20 352
Hotelling-Lawley Trace 0.17 1.40 20 292.24
Roy's Greatest Root 0.13 2.27 10 176
Note: *= p<.05. F Statistic for Wilks' Lambda is exact.
Table 17 presents the descriptive information for the motivation domains of each
grade level. When examining the school performances, there were three clear patterns
that appear. First, there were six domains which showed increases from 3rd to 5th grade
(Efficacy, Challenge, Curiosity, Importance, Competition, and Compliance). Second,
recognition was the only domain that showed a consistent decrease from 3rd to 5th grade.
Third, there were three domains which showed an increase from 3rd to 4th grade and then
a decrease from 4th to 5th (Involvement, Grades, and Social). Overall, the data supported
the notion that motivation is multifaceted and not a singularly represented concept. The
scores ranged from 2.55 (Social) to 3.57 (Importance). Only three domains could be
considered as low motivators for the students with scores below 3.0: Challenge,
Involvement, and Social. Two domains, Grades and Importance, could be considered as
strong areas of motivation for the students with scores above 3.5. The remainder of the
95
95
Pag
e95
2
2 P
age9
52
2
Pag
e95
22
scores fell between 3.0 and 3.5 (Efficacy, Curiosity, Recognition, Competition, and
Compliance).
Similar to the reading data, a MANOVA test was completed to determine if there
were significant overall differences between grade level motivation variables. Table 18
provides the results of the t-tests. There was no evidence that there was an overall
significant difference between the schools for the motivation variables: F(20, 350) = 1.43,
p =0.107.
Motivation Clusters
Similar to what was done previously with the reading factors, a cluster analysis
was completed with the motivation factor scores. The SAS program was used to generate
clusters using both hierarchical and non-hierarchical methods. The average, complete,
single, centroid, and median linkage methods were used to determine the most consistent
clustering structure. First, the dendograms and cluster group history were examined to
determine the number of clusters. The dendograms were examined visually to determine
where the largest distance between clusters occurred. With the large sample size, some
of these were not visually interpretable. The complete and ward’s method visually
showed six clusters.
96
96
Pag
e96
2
2 P
age9
62
2
Pag
e96
22
Table 17
Descriptive Statistics for the Motivation Data by Grade Level
Variable 3rd
n=83
4th
n=54
5th
n=50
Total
n=187
Social 2.49
.68
2.71
.55
2.46
.57
2.55
.62
Challenge 2.72
.74
2.82
.62
2.98
.55
2.82
.67
Involvement 2.76
.64
2.92
.46
2.81
.49
2.82
.56
Compliance 2.97
.48
3.03
.49
3.04
.46
3.00
.48
Competition 3.01
.70
3.09
.64
3.10
.59
3.06
.65
Curiosity 3.04
.63
3.15
.58
3.18
.48
3.11
.58
Efficacy 3.12
.64
3.23
.64
3.29
.55
3.20
.62
Recognition 3.32
.64
3.30
.55
3.21
.60
3.29
.60
Grades 3.43
.62
3.67
.48
3.55
.42
3.53
.54
Importance 3.51
.69
3.57
.61
3.65
.45
3.57
.61
Note: Variables have been sorted with total averages in ascending order.
Possible values can range from 1 to 4. Standard Deviations are italicized.
97
97
Pag
e97
2
2 P
age9
72
2
Pag
e97
22
Table 18
MANOVA Test Results for Motivation Differences Between Grade
Statistic Value F Value Num DF Den DF
Wilks' Lambda 0.85 1.43 20 350
Pillai's Trace 0.15 1.43 20 352
Hotelling-Lawley Trace 0.16 1.42 20 292
Roy's Greatest Root 0.09 1.59 10 176
Note: *= p<.05. Dependent Variable=Grade, Independent Variable(s) = Motivation
Domains.
The cluster history was examined with the same goals while identifying the
cluster solution that would have the largest distance while keeping the smallest RMSSTD
and Semi-partial R-Squared values and the highest R-squared values. This procedure
required the researcher to examine each line, comparing them to the line before and after
to determine where the largest “jumps” occured and use the cluster solution above this.
For this set, the largest jumps occurred at the line for cluster six for semi-partial R-square
and R-Square, this suggested a six cluster solution (see Table 19). A six cluster solution
was selected as the best fit to proceed for non-hierarchical methods. The means from this
cluster solution were used as initial seeds, or starting points, to generate refined clusters
with students who were closest to the initial seeds.
98
98
Pag
e98
2
2 P
age9
82
2
Pag
e98
22
Table 19
Motivation Cluster History
Number
of
Clusters
Clusters Joined n New Cluster
RMS Std
Dev
Semipartial
R-Square
R-Square Between
Cluster
Sum of
Squares
10 CL15 CL24 35 .43 .02 .55 11.32
9 CL12 CL16 28 .51 .02 .53 12.07
8 CL19 CL17 32 .45 .02 .51 14.47
7 CL13 CL14 51 .41 .02 .48 14.88
6 CL9 CL18 36 .55 .03 .46 17.06
5 CL7 CL23 79 .40 .03 .43 19.80
4 CL6 CL11 41 .61 .03 .39 22.26
3 CL10 CL8 67 .48 .04 .36 23.24
2 CL3 CL4 108 .58 .10 .26 64.49
1 CL2 CL5 187 .60 .26 .00 171.83
Note: Ward’s Clustering Method Used. Only possible clusters less than 10 were included
to save space. Shaded line identifies greatest “jumps” and the best cluster.
The k-means non-hierarchical clustering method was used to generate groups.
The Ward’s clustering method was selected as the best option for grouping because it
grouped more similar students together without having clusters with one to two students.
The results produced clusters with varying motivational patterns (see Table 20).
99
99
Pag
e99
2
2 P
age9
92
2
Pag
e99
22
Table 20
Motivation Clusters Summary Data
Cluster
n
Percen
tage
of S
ample
Efficacy
Challen
ge
Curio
sity
Involv
emen
t
Importan
ce
Reco
gnitio
n
Grad
es
Com
petitio
n
Social
Com
plian
ce
1 40 21.4% 3.88 3.56 3.62 3.16 3.95 3.71 3.92 3.73 3.05 3.17
2 39 20.9% 3.32 3.06 3.44 3.12 3.85 3.53 3.74 3.10 2.79 3.08
3 37 19.8% 2.80 2.42 3.09 2.99 3.41 3.31 3.46 2.94 2.72 3.03
4 35 18.7% 3.31 2.71 2.87 2.35 3.73 3.35 3.51 3.09 1.90 3.12
5 31 16.6% 2.63 2.38 2.49 2.51 2.87 2.54 3.10 2.45 2.27 2.65
6 5 2.6% 2.53 1.44 1.97 1.73 2.70 1.96 2.05 1.80 1.54 2.28
When looking at each cluster, there were unique characteristics present. First, no
cluster contained more than 25% of the sample. This meant there were not groups that
were overtly large with the most similar patterns. Second, there was no single
consistently high motivator across the six groups. The domains of importance and grades
were the highest motivator between the six groups.
A closer examination of the clusters provided a clearer interpretation. Clusters
one and six, respectively, represent the highest motivated students and the lowest
motivated students. The highest motivated students in cluster one represented the largest
cluster. They have relatively the highest scores for each motivational domain. The five
students in cluster six students have the lowest motivation scores for each motivational
100
10
0
Pag
e10
0
2
2 P
age1
00
22
Pag
e10
02
2
domain. Clusters two and five have students with consistently high or average
motivators. Cluster two students have consistently high scores with the exception of their
lack of participation in social related reading activities. The students in cluster five are
average with most scores between 2.0 and 3.0 with the exception of Grades, which is
above 3.0. Last, the students in clusters three and four are characterized as “mixed
motivated.” They have combinations of high and low motivators, and no noticeable high
or low areas.
Motivation and Reading Clusters
Motivation and reading factors were previously identified within this dissertation.
The exploratory factor analysis identified two reading factors: word identification and
meaning. The confirmatory factor analysis confirmed ten motivation factors: Efficacy,
Challenge, Curiosity, Involvement, Importance, Recognition, Grades, Competition,
Social, and Compliance. Earlier, the cluster analysis procedure identified four clusters
that represented the reading factors, and six that represented the motivational factors.
While these analyses provided important information about the patterns of students, the
information is limited to grouping based on either motivation or reading. Previous
researchers (Baker & Wigfield, 1999; Guthrie, Coddington, & Wigfield, 2009) used
motivation to group students into clusters, then used summary statistics to describe the
patterns of performance for their sample. While this is acceptable, a grouping that takes
into account both the reading and motivational factors generates clusters that are
representative of students’ patterns of performance with both areas. This study had a
goal to better understand non-proficient students by grouping the students based on both
101
10
1
Pag
e10
1
2
2 P
age1
01
22
Pag
e10
12
2
their motivation and reading descriptors using the clustering process. I use this section to
present results of the cluster analysis, including both reading and motivation factors
combined, to understand the different patterns of reading and motivation that exist
amongst non-proficient minority students. The process is explained, followed by a brief
description of the results of the process.
Although the number of clusters from the reading and motivation data were
identified separately, this is not representative of the true number of clusters when
including all factors together. Therefore, the complete process of using hierarchical and
non-hierarchical clustering techniques was completed to determine the number of groups
and define the membership (Sharma, 1996). As with the previous cluster analyses, the
clustering process was completed with the SAS program using both hierarchical and non-
hierarchical methods including analyses using all of the five linkage methods. See above
sections for the specific steps followed for this analysis.
The cluster history was examined with the same goals, while identifying the
cluster solution that would have the largest distance while keeping the smallest RMSSTD
and Semi-partial R-Squared values and the highest R-squared values. For this set, the
largest jumps occurred at the line for cluster six for Semi-partial R-square and R-Square,
suggesting a six cluster solution (see Table 21). Similar to the motivation cluster
analysis, a six cluster solution was selected as the best fit to proceed for non-hierarchical
methods. The means from this cluster solution were used as initial seeds, or starting
points, to generate refined clusters with students who were closest to the initial seeds.
102
10
2
Pag
e10
2
2
2 P
age1
02
22
Pag
e10
22
2
Table 21
Combined Cluster History
Number
of
Clusters
Clusters Joined n New Cluster
RMS Std Dev
Semipartial
R-Square
R-Square Between
Cluster
Sum of
Squares
10 CL41 CL23 14 .54 .01 .48 4.23
9 CL14 CL34 35 .50 .03 .45 4.31
8 CL11 CL21 18 .63 .01 .44 4.41
7 CL18 CL8 37 .62 .03 .41 4.60
6 CL16 CL9 46 .58 .04 .37 5.24
5 CL6 CL10 60 .65 .08 .29 5.95
4 CL7 65 38 .65 .02 .27 6.17
3 CL5 CL12 148 .61 .08 .19 6.21
2 CL4 38 39 .69 .03 .16 7.56
1 CL3 CL2 187 .68 .16 .00 8.61
Note: Ward’s Clustering Method Used. Only possible clusters less than 10 were included
to save space. Shaded line identifies greatest “jumps” and the best cluster.
With the number of clusters determined, the non-hierarchical process was used to
refine the membership in each of the six groups. This process was completed by using the
means from each cluster as the initial seeds (starting points) for the refining process. This
process grouped students around the starting values, which ensured students were close to
others with similar scores. The revised six cluster solution produced clusters with varying
103
10
3
Pag
e10
3
2
2 P
age1
03
22
Pag
e10
32
2
frequencies and distances from other clusters. Table 22 identifies the frequencies, along
with the summary data for each factor.
Table 22
Combined Clusters Summary Data
Cluster 1 2 3 4 5 6
Frequency 30 25 48 19 44 21
Compliance 3.13 3.07 2.91 2.97 3.20 2.58
Social 2.82 2.21 2.35 2.44 3.03 2.07
Competition 3.64 2.66 3.01 3.10 3.31 2.24
Grades 3.89 3.41 3.61 3.22 3.81 2.65
Recognition 3.65 2.84 3.30 3.27 3.70 2.41
Importance 3.92 3.42 3.64 3.24 3.85 2.79
Involvement 3.18 2.72 2.72 2.75 3.08 2.17
Curiosity 3.68 2.85 2.98 3.00 3.40 2.36
Challenge 3.52 2.76 2.55 2.56 3.15 2.06
Efficacy 3.83 3.16 3.11 2.86 3.37 2.51
Meaning .77 1.55 -.61 .04 -.50 -.53
Word ID .63 .29 .57 -2.08 -.21 -.22
Each profile is unique and the patterns present were used to better understand
each group of students. Three unique patterns are present within the data. First, through
all of the clusters, the motivation domains of Grades and Importance of reading were
104
10
4
Pag
e10
4
2
2 P
age1
04
22
Pag
e10
42
2
consistently the highest motivators for reading. Because these are high for the majority of
the sample, they will not be used to describe and compare the clusters later. Second, the
reading data within the clusters have combinations of high and low with scores above and
below the median. Third, the social domain has the lowest scores for all the clusters, with
the exception of cluster six, which comes close. Last, the students in cluster one have the
highest motivating areas for reading and the highest word recognition scores, but do not
have the highest meaning scores. The paragraph provides descriptive information to
differentiate the motivation for reading and reading performances between the clusters.
Each cluster has unique characteristics that help with better understanding these
non-proficient readers. Clusters one and five contained the students with the highest
motivation for reading with consistently high scores for each motivation domain in
comparison to the other clusters. The students in cluster one have high word
identification and meaning scores. The students in cluster five have low word
identification and meaning scores as they fell just below the median. The students in
clusters two and three have higher word identification scores; however, their meaning
abilities differed. Cluster three students have the lowest meaning scores while cluster
two have the highest meaning scores. When examining their motivation for reading,
Grades and Efficacy drive these students, while Challenge and Involvement do not
necessarily drive these students to participate in reading related activities. Cluster four
students have the lowest word recognition, while having average meaning scores. These
students’ motivation for reading have strength in their Curiosity about reading and their
105
10
5
Pag
e10
5
2
2 P
age1
05
22
Pag
e10
52
2
competitive nature. Last, cluster six students have the lowest scores in all motivation to
read areas. Grades are their highest motivator to participate in reading activities.
This chapter presented the results of the research study. Through the various
analysis, the results highlighted the differences that exist between non-proficient students.
These students overall are performing below grade level in all areas and have significant
differences between their comprehension of narrative and expository texts. Using factor
analysis, two factors were found that represent the reading data; word identification and
meaning. Ten of the motivation for reading domains were identified using factor analysis
of the motivation data set; however, the domain of avoidance was not present. These
results provide support for three major ideas. First, there is heterogeneity in the academic
performance and motivation for reading of non-proficient readers. Second, the idea that a
one-size-fits-all model does not work for providing remedial instruction for non-
proficient readers (Allington, 2009). Last, the results support the need to use motivation
when addressing the academic needs of non-proficient students.
106
10
6
Pag
e10
6
2
2 P
age1
06
22
Pag
e10
62
2
CHAPTER V
DISCUSSION
The purpose of this study was to examine the reading and motivational profiles
for students who have been identified as non-proficient readers. Accountability policies
tend to unfairly penalize those schools with high percentages of marginalized groups,
students of color and those living in poverty (Hursh, 2007). Because these schools often
receive low-performing or failing designations, research needs to include such students in
their studies if we want to understand their reading strengths and weaknesses. Au (2009)
described the “zip code effect,” where schools within geographical areas tend to represent
a specific socio-economic status and ethnic makeup. This study purposefully included
zip code schools to ensure the representation of these marginalized students, who often
are criticized for being “unmotivated” and/or low achievers. Moreover, these schools
traditionally focus solely on word identification strategies when providing remediation. If
these strategies do not address students’ overall needs, then schools need to use a variety
of other assessments and interventions to address the heterogeneity that exists amongst
these students.
This chapter first discusses individual reading and motivational profiles. Within
this discussion, I focus mainly on how the profiles from the separate analyses in this
study differed from existing studies where such profiles also were generated. Second, I
place the larger emphasis of this chapter on a discussion of the combined reading and
107
10
7
Pag
e10
7
2
2 P
age1
07
22
Pag
e10
72
2
motivational profiles as these profiles were the goal of the study. Within this chapter, I
highlight similarities and differences between the separated and combined clusters. Last,
I share implications for practice and policy, followed by suggestions for the need to
conduct future research.
Reading Profiles
Researchers have presented different profiles of readers (Buly & Valencia, 2002;
Dennis, 2013; Leach et. al, 2003; Leseaux & Kieffer, 2010; Meyer et. al, 2013; Rupp &
Leseaux, 2006; Pierce et. al, 2007). When identifying such profiles, only two studies
placed an emphasis on non-proficient readers and used assessments measuring the
National Reading Panels’ (NICHD, 2000) five components of reading (Buly & Valencia,
2002; Dennis, 2013). Both studies identified a three-factor representation of word
recognition (decoding), fluency, and meaning. Both studies used the three factors to
generate from four to 10 profiles.
In the reading only cluster analysis of this study, two factors emerged, word
identification and meaning. Thus, unlike most of the previous studies using exploratory
factor analysis, except for the Rupp and Leseaux (2006) study, fluency did not appear as
a separate factor. The two factors, word identification and meaning, were used in a
cluster analysis and generated four reading profiles. There are three important things to
note from these four profiles. First, unlike most existing research of reading profiles
(Buly & Valencia, 2002; (Leach et. Al, 2003; Leseaux & Kieffer, 2010; Meyer et. Al,
2013; Rupp & Leseaux, 2006; & Pierce et. al, 2007), this study’s separate analysis failed
to identify a group of non-proficient readers who had a combination of low scores on the
108
10
8
Pag
e10
8
2
2 P
age1
08
22
Pag
e10
82
2
word identification and meaning factors. This difference mainly related to students’
word recognition abilities, which were clearly somewhat stronger in this study than in
previous related studies. An important note, however, the subsequent combined analyses
did reveal such a pattern, but only represented a small portion of the sample. Second, no
single set of profiles represented all non-proficient readers (Buly & Valencia, 2002);
differences in profiles are a direct reflection of the sample analyzed. This finding runs
counter to present accountability interventions (Allington, 2009) where non-proficient
readers are presented with one-size-fits-all remedial instruction. Third, overall, students
in each cluster scored higher on their ability to decode or recognize words than on their
comprehension of texts. Moreover, differences of almost a grade level existed between
students’ ability to comprehend narrative versus expository texts. Other profile specific
similarities and differences will be presented when the results from the combined analysis
are discussed.
Motivation Profiles
Motivation for reading explains why students approach or avoid reading
activities. While researchers have explored students’ motivation for reading, their
samples have not emphasized non-proficient readers. Based on its existing use with
elementary students and its range of motivation domains, the MRQ was used to generate
motivational profiles. Existing studies have identified six to 11 factors. This study’s
confirmatory factor analysis identified 10 of the 11 original factors. These included
Challenge, Competition, Compliance, Curiosity, Efficacy, Grades, Recognition, Social,
Importance and Involvement. Unique to this study, an avoidance factor was not
109
10
9
Pag
e10
9
2
2 P
age1
09
22
Pag
e10
92
2
discovered. This is important because it documents students’ positive orientations
towards reading in comparison to some general expectations for this group of readers. In
other studies, avoidance was a common factor (Baker & Wigfield, 1999; Unrau &
Schlackman, 2006; Wigfield & Guthrie, 1997). Moreover, an avoidance factor is a
central component to the expectation-value model, upon which the MRQ is based. More
emphasis will be directed towards this point when discussing the combined analysis.
The 10 motivation factors were used in the clustering process to generate
profiles. This analysis produced a six-cluster solution. The cluster analysis produced six
profiles, which was fewer than the seven profiles in the work of Baker and Wigfield
(1999). These differences in the number of profiles were a clear reflection of the sample
used, in that, this study’s sample included greater heterogeneity. Subsequent discussion
of these differences is presented with the analysis of the combined clusters.
Motivation and Reading Profiles
This study’s main purpose was to evaluate reading and motivation profiles in a
combined analysis for non-proficient readers. The analysis specifically addressed
marginalized groups, including high percentages of ethnic minority students whose
families lacked economic resources. Prior studies have failed to directly address such
student populations, nor have they simultaneously analyzed reading and motivation
factors represented for non-proficient readers.
To better understand these students, combined reading and motivation profiles
were generated (see Table 29). The motivational domains of Grades and Importance
received the highest ratings across all profiles, whereas Social domain had the lowest
110
11
0
Pag
e11
0
2
2 P
age1
10
22
Pag
e11
02
2
ratings. As expected, the patterns of the ratings differed across the six profiles. When
interpreting these profiles, I used their reading performances as a basis to describe the
findings because reading proficiency was the main construct of the study. The profiles
were named using descriptive terminology that will allow teachers to identify students’
placement using the diagnostic reading behaviors of students as well as their motivation
for reading characteristics.
First, while two of the groups had reading performances above the mean for this
sample, their motivational orientations differed markedly. While the scores were above
the mean, these students are non-proficient and not meeting grade level expectations.
Students in the first profile, Motivated Readers (n=30), performed higher on meaning,
yet lower on word recognition than students in the second profile, and had above average
ratings on each of the motivation indices. In particular, these students had the highest
scores on Importance, Curiosity, Recognition, and Competition. Students in the second
profile, Confident and Compliant Readers (n=25), performed above average on both
meaning and word recognition measures, yet had average motivation scores for seven of
the motivation measures, with below the mean scores for Social, Competition, and
Curiosity. The students in both profiles scored above the mean for word identification
and meaning, the difference between these profiles is attributed to the motivation scores
being higher for profile one than profile two.
The next two profiles had combinations of the lowest scores for one of the reading
factors. Students in profile 3, Competitive Word Callers (n=48), performed above
average on word identification, but had the lowest scores for meaning in the sample, and
111
11
1
Pag
e11
1
2
2 P
age1
11
22
Pag
e11
12
2
had average ratings for eight of the motivation indices with Challenge and Social having
below average ratings. Profile 4 students, Curious Comprehending Readers (n=19),
performed at the mean for meaning, but had the lowest word identification scores in the
sample. Six of the motivation domains were average for these students with below the
mean ratings for grades, importance, challenge, and efficacy. These two groups had
similar patterns of high and low motivation areas; however, the main difference here is
their reading performances. The students in profile three are “word callers”, who are not
making meaning, while the students in profile four are struggling with their word
identification skills.
The final two profiles had lower than the mean scores for both reading factors.
Both clusters had almost identical scores for both word identification and meaning. The
differences between these students once again was attributed to the motivation domains.
Profile five students, Motivated Multi-Need Readers (n=44), had four motivation ratings
above the mean in Grades, Recognition, Importance, and Social with the other motivators
falling below the mean. The students in profile six, Compliant Multi-Need Readers
(n=21), had scores below the mean for each of the motivators. Although they are below
the mean, they still had a preferred motivator, being compliant. The difference between
these two profiles is attributed to motivation. The students in profile five are low
performing but have high motivation scores, while the students in profile six have low
reading performance patterns and low motivation scores.
112
11
2
Pag
e11
2
2
2 P
age1
12
22
Pag
e11
22
2
Differences Between Conducting a Separate Versus a Combined Analysis
There are three clear similarities and differences to note between the reading
profiles and the combined profiles. First, the number of profiles were fewer when the
reading factors were evaluated separately. There are two reasons for this finding.
Statistically, we would expect fewer profiles due to the number of considered variables.
Next, the motivation variables allowed for further discriminations when considering
students’ performances across the two domains. Second, the separate analysis of the
reading clusters did not identify a group of students who had below the mean scores for
both word identification and meaning. The combined cluster analysis identified two
groups that had a combination of below mean scores for both word identification and
meaning. The original four patterns were present within clusters one through four;
however, by adding a motivational component, students who were low in both reading
areas were separated into two unique groups. Last, each of the largest profiles in both
analysis had what Buly and Valencia (2002) term “word callers.” These students had
stronger word identification scores and weaker meaning as they were unable to make
meaning at the same level that they could decode.
When examining differences between the separate and combined analysis for
motivation, there were four similarities and differences to note. First, the separate
motivation analysis produced a profile with only five students. This is quite a small
group considering the sample size. In comparison, the smallest profile in the combined
analysis was 19 students. Second, in the combined clusters, importance and grades were
consistently the highest motivators for each cluster. In the motivation only analysis,
113
11
3
Pag
e11
3
2
2 P
age1
13
22
Pag
e11
32
2
profile five had recognition as the highest motivator. Third, both sets of clusters
contained a profile with the highest scores for each motivation domain and another
profile with students with the lowest scores. There are two possible explanations for
these differences. Statistically, when one increases the number of variables, greater
variation exists. Next, due to the nature of the variables across the two domains of
reading and motivation, greater differentiation resulted. These explanations might have
operated together to produce these differences.
In summary, these profiles demonstrated a need to include multiple variables,
particularly those related to motivation, to better understand non-proficient students.
When examining the patterns present, it was essential to understand the reading strengths
and weaknesses as well as identifying the most versus less preferred motivators. Instead
of classifying non-proficient and proficient readers as “motivated” or “not motivated,” we
should adopt a more multi-dimensional perspective by evaluating the multiple reasons as
to why students are or are not motivated to engage in a particular task (Baker & Wigfied,
1999). That is, motivation is multi-dimensional: it consists of patterns of underlying
motivational constructs that guide students’ participation in reading. These profiles
consisted of patterns with some students having preferences that motivate them to
participate in reading related activities and other areas that are not as desired. Also, the
identification of motivation in addition to the reading patterns is essential because the low
performing readers were not represented solely with the lower motivations for reading.
These lower performing students are often considered by teachers as not motivated when
114
11
4
Pag
e11
4
2
2 P
age1
14
22
Pag
e11
42
2
they have differing motivators. To the contrary of existing work, with this “zipcode”
population, they did not demonstrate an avoidance performance orientation.
The profiles demonstrated that non-proficient readers from marginalized
populations were not a homogenized group (Dennis, 2013). The profiles consisted of
various numbers of students, made up of various ethnic groups. No one ethnic group was
represented as the dominant population in any of the profiles. Regardless of the profile,
they have unique needs that must be addressed via assessments with interventions to
build on their strengths while addressing their weaknesses. A one-size-fits-all
instructional approach does not benefit these students because it fails to place a direct
emphasis on the areas of strength and weakness.
Finally, when considering motivation, it is vital to dig deeper than just
acknowledging domains which are preferred or less preferred for students. In this study,
the motivation areas of Importance, Grades, and Social had consistent patterns across the
various profiles. As such, only interpreting these areas as part of the profiles would have
caused all profiles to be noted with almost identical motivational patterns. It is important
to take into consideration the overall domains with higher and lower preference and
understand how it impacts students’ total performances.
Implications
Informing Practice
Differentiated instruction is a key component for these non-proficient students.
Identification of a student as non-proficient via a standardized assessment does not
adequately describe the many facets of a student’s reading or motivational performances.
115
11
5
Pag
e11
5
2
2 P
age1
15
22
Pag
e11
52
2
The main point is that non-proficient students fall into multiple patterns. Teachers must
use assessments that measure beyond the word identification components of reading.
These assessments will help teachers to identify specific patterns of reading strengths and
weaknesses for these students. The patterns can then be used to identify specific areas for
interventions to assist with remediating areas of weakness and ensuring the existing
achievement gaps are closed.
The results of the study support the need to understand and identify the
motivation for reading for non-proficient students. Some students within a classroom
may appear as low in all reading skills, however; they may not have low motivation
preferences across all domains. There will be clear preferences across their motivational
orientations. When teaching non-proficient students, these areas can be addressed as
motivators to encourage student participation in reading activities. In this study, social
reading was a low preference across all profiles. The students in this study would not
benefit from tasks that require them to read with a family member or complete specific
reading related activities outside of class (i.e., visiting the library, helping friends with
their schoolwork, telling their family and friends about what they are reading). Of
course, this recommendation might change after students improve their reading.
However, with the clear high scores for importance of reading, these students would
benefit from the teacher emphasizing why particular activities are of importance. While
multiple motivation surveys exist, surveys used should measure multiple motivation
domains. This would ensure that teachers adequately identify multiple motivators present
with their students.
116
11
6
Pag
e11
6
2
2 P
age1
16
22
Pag
e11
62
2
These profiles present specific differentiations that can be used for students who
are represented within each profile. In differentiating, teachers should place emphasis on
the reading weaknesses of students, as this is the cause of students reading failures, while
also engaging students in reading activities based on their motivation domains with the
highest scores. In describing implications for differentiating instruction for each of the
profiles, I looked to those motivation factors which students in a profile rated most highly
as areas upon which teachers could focus their efforts to increase engagement. The
students who are described as Motivated Readers will benefit from interventions devoted
to helping them make meaning when they read. Regarding motivation, they uniquely had
high scores on 9 of the 10 items, giving teachers multiple ways to promote their
engagement. They particularly rated Grades, Importance, Curiosity and Efficacy as high
areas; therefore, teachers could appeal to these areas to promote their engagement. They
also could use any of the other five areas as an approach. These interventions and
activities should center on things of interest to the students, and things that will peak their
curiosity and help to increase their engagement while working on the students making
meaning (Vansteenkiste, Lens, & Deci, 2006). The Confident and Compliant Readers
believe they are decent readers, but they are also compliant, completing reading activities
because of external requirements. Regarding motivation, their four highest ratings were
Grades, Importance, Efficacy and Compliance factors. Interventions for these students
should emphasize fluency to increase the amount of words read per minute, while
continuing to reinforce making meaning. These interventions should encourage students
to continue to have strong beliefs of themselves as good readers meeting comprehension
117
11
7
Pag
e11
7
2
2 P
age1
17
22
Pag
e11
72
2
proficiency, but need to read with increased speed and accuracy (Wigfield, Guthrie,
Tonks, & Perencevich, 2004).
The Competitive Word Callers will benefit from interventions focused on making
meaning. These students had the highest scores for grades, importance, efficacy and
competition and a strength in word recognition. These students must continue to hear
and believe they are good readers (Wigfield et. al., 2004), but with the emphasis of being
a good reader shifting from word identification to making meaning. With their
competitive nature, their interventions on making meaning should include charts or lists
that acknowledge their continued improvement in making meaning.
The Curious and Competitive Comprehending Readers need an emphasis placed
on their word recognition skills while still addressing meaning. These students had the
highest scores for Grades, Importance, Recognition, and Competition. Comprehension
(making meaning) is the ultimate goal of reading and although interventions for this
group should emphasize decoding and/or fluency, they should additionally address the
students making meaning. With the high desire to be recognized for their reading
behaviors, these students should be recognized for accomplishments through a goal chart
or a similar activity that allows the teacher to celebrate and recognize these student’s
success.
The Motivated Multi-Need Readers need interventions emphasizing both word
recognition and meaning, as both are weaknesses. Regarding motivation, similar to
profile one, they uniquely had high scores on all ten of the motivation items, giving
teachers multiple ways to promote their engagement. They particularly rated Grades,
118
11
8
Pag
e11
8
2
2 P
age1
18
22
Pag
e11
82
2
Importance, Curiosity and Recognition; therefore, teachers could address these areas to
promote reading engagement. Interventions for this profile could be completed by
introducing and reinforcing decoding strategies during intervention time. While
addressing these strategies, the students would also benefit from explicit teaching of
comprehension related strategies through a variety of texts read by the students, with the
students and to the students. The interventions and activities for this group should
encourage the curiosity of students as they explore the reading by connecting to topics
and subjects with high interest and using highly engaging strategies (Meece, Anderman,
& Anderman, 2006).
The Compliant Multi-Need Readers represent the most challenging of the profiles.
These students have low scores for word recognition and meaning, and after Grades and
Importance, their highest motivational ratings are for Compliance. A teacher would not
want to encourage this set of behaviors by themselves; instead, they should focus on
promoting their curiosity, which has a closer rating, as a means to encourage their
engagement. Interventions for students in this cluster should emphasize meaning first,
while also addressing word recognition skills. Interventions and activities for this group
should encourage engagement by addressing their curiosity through high interest reading
texts and hands-on activities (Guthrie et al., 2007).
Informing Policy
As federal and state governments continue to enact accountability legislation, they
must consider how and what interventions will be implemented to meet these
accountability requirements. Instead of promoting one-size-fits-all remediation
119
11
9
Pag
e11
9
2
2 P
age1
19
22
Pag
e11
92
2
(Allington, 2009), federal funding should be used to support the identification of
differentiation strategies with non-proficient readers. As emphasized in the implications
for practice section, these strategies should not be limited to programs, but should be
inclusive of actual reading and motivation strategies that can be shared with teachers to
use in their classrooms to differentiate instruction. With the low proficiency of
marginalized groups, they must be a direct beneficiary of new interventions. The
interventions must consider who will benefit from the strategy, with the emphasis placed
on non-proficient students.
As mandates continue to be made with regards to education, policy makers need
to make realistic considerations. The results of this study demonstrate the heterogeneity
that exists among non-proficient students. With this range of patterns, it is not realistic to
expect all students to meet grade level expectations immediately. Mandates for
proficiency should take this into consideration. Proficiency models should also consider
growth as an accountability factor. These models should take into account growth of
individual reading skills that are not measured by a single end-of-grade assessment.
Last, policy makers on local levels should continue to encourage differentiation
and equitable interventions for each reader. This differentiation for reading could occur
in two ways. First, schools can continue to require an intervention and enrichment block.
This time could be used to provide direct needs-based instruction that is related to the
students’ weaknesses (Averill, Baker, & Rindaldi, 2014). Second, schools can continue
to require a guided reading block. Guided reading is a part of a balanced literacy
framework that provides small group explicit instruction rooted in the individual needs of
120
12
0
Pag
e12
0
2
2 P
age1
20
22
Pag
e12
02
2
students (Fawson & Reutzel, 2000; Fountas & Pinnell, 2012). This guided reading time
is beneficial for students to practice a variety of reading related skills daily with text on
their instructional level. This allows non-proficient students like those in this study to
receive instruction that addresses their weakness on a daily basis.
Future Work
From this study, multiple follow-up studies could occur. I will discuss four
possible avenues. First, in this sample of non-proficient readers, the motivation domain
of avoidance was not represented in the data. The sample size of the data set was close to
the acceptable range, but realistically it fell just slightly below the 200-400 range. Other
studies identified similar factor structures, but the MRQ structural model should be
examined using a larger sample, ideally 300-400 non-proficient students. This will
determine if there are consistent motivation domains represented across multiple samples.
Second, the motivation and reading profiles should be studied longitudinally
along with interventions based on their motivational preferences and their reading
strengths and weaknesses. Are memberships in profiles changing as would be desired, or
are they showing consistent increases in multiple areas and maintaining the same
patterns? While reading profiles are useful for understanding the patterns of non-
proficient students, the ultimate goal is reading proficiency. As such, we ideally want the
students’ patterns to change as they continue to grow as readers and hopefully achieve
proficiency. Are these patterns changing? Do components change? How do the changes
occur? These are all questions that could be investigated in a longitudinal study.
121
12
1
Pag
e12
1
2
2 P
age1
21
22
Pag
e12
12
2
Third, fluency was not identified as a separate factor in this study. Theoretically
this combined factor including rate and accuracy would be expected with students who
are exposed to interventions and instruction focused on word recognition. Perhaps this
finding is related to the types of interventions offered to students, where the emphasis
was on the decoding of individual words with minimal focus on the reading of connected
text. Is this the case with other non-proficient students? Future work should investigate a
larger number of reading variables within the five areas of reading to determine if these
two areas will be combined or separate into isolated factors.
Fourth, the descriptive data noted at least one grade level difference between the
comprehension of narrative and expository texts for this sample. With the
implementation of Common Core State Standards, there has been a push to increase
students’ experiences with expository texts. There is a need to determine if the recent
emphasis on expository texts has resulted in increased comprehension of expository texts.
With this, we must determine whether or not this pattern is unique to this sample or is a
common trend amongst all non-proficient readers.
Limitations
Limitations exist with the current study, although I attempted to avoid many of
these. First, this study contained a specific sample not specifically present within all
schools. The sample was non-proficient students, specifically those who are from
marginalized (ethnic and economic) groups. With this, the results of this study with
regards to factors may not be generalizable to all non-proficient groups, nor to proficient
students. Second, the largest possible sample representing “zip code” schools was used
122
12
2
Pag
e12
2
2
2 P
age1
22
22
Pag
e12
22
2
in the study. A confirmatory factor analysis was used to analyze the fit of the MRQ
domains. While my sample came close, it was not within the desired range of 200-400
for a CFA. Future work should seek to identify and use a larger sample to confirm the
lack of representation of the avoidance factor. Last, in terms of evaluating
comprehension, students’ background knowledge impacts their ability to make meaning
from text. The passages used in this study were selected by staff at the schools so that all
students should have had background knowledge of the topics. However, there may have
been passages for which the students had little or no background knowledge, which
would impact their ability to connect and make meaning.
Conclusions
These results are beneficial to parents, teachers, administrators, and policy
makers. There is no one-size-fits-all approach to remediating non-proficient readers.
These readers have multiple patterns of strengths and weaknesses. In addition to their
patterns of reading strengths and weaknesses, these students also have multiple patterns
of motivation preferences. While this study identified six profiles representing reading
and motivation patterns, the number of profiles are not important. The patterns highlight
the need to dig deeper and examine the motivation preferences, and reading strengths and
weaknesses. The designation of “non-proficient” from external assessments does not
provide information necessary to provide a direct intervention to improve the academic
outcomes for these students. This designation should merely be an identifier that gives a
license to dig deeper through multiple assessments in order to understand what
specifically needs to be addressed. This assessment data can be used to provide
123
12
3
Pag
e12
3
2
2 P
age1
23
22
Pag
e12
32
2
meaningful differentiation for these non-proficient students. Through differentiation, the
ultimate goal of student growth and proficiency may be achieved.
124
12
4
Pag
e12
4
2
2 P
age1
24
22
Pag
e12
42
2
REFERENCES
Afflerbach, P. (2004). Assessing adolescent reading. In T. L. Jetton & J. A. Dole (Eds.),
Adolescent literacy research and practice (pp. 369-391). New York: Guilford.
Allington, R. L. (2001). What really matters for struggling readers. New York:
Longman.
Allington, R. L. (2009). Literacy policies that are needed: Thinking beyond No Child
Left Behind. In Y. Goodman & J. Hoffman (Eds.), Changing literacies, changing
times: A historical perspective on the future of reading research, public policy,
and classroom practices. (pp. 266-281). New York: Routledge.
Allington, R. L., & McGill-Franzen, A. (1992). Unintended effects of educational
reform in New York. Educational Policy, 6, 397-414.
Amrein, A. L. & Berliner, D. C. (2003). The effects of high states testing on student
motivation and learning. Educational Leadership, 60, 32-38.
Atkinson, J. W., & Feather, N. T. (Eds.). (1966). A theory of achievement
motivation (Vol. 66). New York: Wiley.
Au, W. (2009). Unequal by design: High-stakes testing and the standardization of
inequality. London: Routledge.
Averill, O.H., Baker, D., & Rinaldi, C. (2014). A blueprint for effectively using RTI
intervention block time. Intervention in School and Clinic, 50, 29-38.
125
12
5
Pag
e12
5
2
2 P
age1
25
22
Pag
e12
52
2
Baker, L., & Wigfield, A. (1999). Dimensions of children’s motivation for reading and
their relations to reading activity and reading achievement. Reading Research
Quarterly, 34, 452–477.
Brown-Jeffy, S., & Cooper, J. E. (2011). Toward a conceptual framework of culturally
relevant pedagogy: An overview of the conceptual and theoretical
literature. Teacher Education Quarterly, 38, 65-84.
Buly, M. R., & Valencia, S. W. (2002). Below the bar: Profiles of students who fail state
reading assessments. Educational Evaluation and Policy Analysis, 24, 219-239.
Burchinal, M., McCartney, K., Steinberg, L., Crosnoe, R., Friedman, S. L., McLoyd., V.,
& Pianta, R. (2011). Examining the Black-White achievement gap among low-
income children using the NICHD study of Early Child Care and Youth
Development. Child Development, 82, 1404-1420. doi:10.1111/j.1467-
8624.2011.01620.x
Cambria, J. & Guthrie, J.T. (2010). Motivating and engaging students in reading. New
England Reading Association Journal, 46, 16-29.
Cattell, R. B. (1966). "The scree test for the number of factors." Multivariate Behavioral
Research, 1, 245-276.
Chapman, J. W., & Tunmer, W. E. (1995). Development of young children's reading self-
concepts: An examination of emerging subcomponents and their relationship with
reading achievement. Journal of Educational Psychology, 87, 154.
Clotfelter, C. T., Ladd, H. F., & Vigdor, J. L. (2009). The academic achievement gap in
grades 3 to 8. The Review of Economics and Statistics, 91, 398-419.
126
12
6
Pag
e12
6
2
2 P
age1
26
22
Pag
e12
62
2
Common Core State Standards Initiative. (2010). Common core state standards for ELA.
Retrieved from http://corestandards.org.
Costello, A. B. & Osborne, J. W. (2005). Exploratory Factor Analysis: Four
recommendations for getting the most from your analysis. Practical Assessment
Research & Evaluation, 10, 1-9.
Crowley, S. L., & Fan, X. (1997). Structural equation modeling: Basic concepts and
applications in personality assessment research. Journal of Personality
Assessment, 68, 508-531.
Dennis, D. V. (2013). Heterogeneity or homogeneity: What assessment data reveal about
struggling adolescent readers. Journal of Literacy Research, 45, 3-21.
Dunn, L. M., & Dunn, L. M. (2007). Peabody Picture Vocabulary Test (4th ed.). Circle
Pines, MN: American Guidance Service.
Eccles J. S., Adler, T. F., Futterman, R., Goff, S. B., Kaczala, C. M., Meece, J. L., &
Midgley, C. (1983). Expectancies, values, and academic behaviors. In J. T.
Spence (Ed.), Achievement and achievement motivation (pp. 75–146). San
Francisco, CA: W. H. Freeman.
Eccles, J.S., Wigfield, A., & Schiefele, U. (1998). Motivation to succeed. In N.
Eisenberg (Ed.), Handbook of child psychology (Vol. IV, pp. 1017–1095). New
York: John Wiley.
Elmore, R. F. (2002) Bridging the gap between standards and achievement: The
imperative for professional development in education. Washington, DC: Albert
Shanker Institute.
127
12
7
Pag
e12
7
2
2 P
age1
27
22
Pag
e12
72
2
Fawson, P., & Reutzel, D. (2000). But I only have a basal: Implementing guided reading
in the early grades. The Reading Teacher, 54, 84-97.
Fisher, D., & Ivey, G. (2006). Evaluating the interventions for struggling adolescent
readers. Journal of Adolescent and Adult Literacy, 50, 180-189.
Fountas, I., & Pinnell, G. (2012). Guided reading: The romance and the reality. The
Reading Teacher, 66, 268-284.
Gaddis S. M., Lauen D. L. (2014). School accountability and the Black-White test score
gap. Social Science Research, 44, 15–31.
Gambrell, L. B., Palmer, B. M., Codling, R. M., & Mazzoni, S. A. (1996). Assessing
motivation to read. The Reading Teacher, 49, 518-533.
Good, R. H., & Kaminski, R. A. (Eds.). (2002). Dynamic indicators of early basic literacy
skills (6th ed.). Eugene, OR: Institute for the Development of Educational
Achievement.
Good, R. H., & Kaminski, R. A. (2011). DIBELS next assessment manual. Longmont,
CO: Sopris.
Graham, S., & Taylor, A. Z. (2002). Ethnicity, gender, and the development of
achievement values. In A. Wigfield, J. Eccles (Eds.), Development of
Achievement Motivation: A Volume in the Educational Psychology Series, (pp.
121-146). San Diego, CA: Academic Press.
Guthrie, J. T., Hoa, A. L. W., Wigfield, A., Tonks, S. M., Humenick, N. M., & Littles, E.
(2007). Reading motivation and reading comprehension growth in the later
elementary years. Contemporary Educational Psychology, 32, 282-313.
128
12
8
Pag
e12
8
2
2 P
age1
28
22
Pag
e12
82
2
Guthrie, J.T, Coddington, C.S., & Wigfield, A. (2009). Profiles of motivation for reading
among African American and Caucasian students. Journal of Literacy Research,
41, 317-35.
Heckhausen, H. (1977). Achievement motivation and its constructs: A cognitive
model. Motivation and emotion, 1, 283-329.
Henson, R. K., & Roberts, J. K. (2006). Use of exploratory factor analysis in published
research common errors and some comment on improved practice. Educational
and Psychological measurement, 66, 393-416.
Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure
analysis: Conventional criteria versus new alternatives. Structural equation
modeling: a multidisciplinary journal, 6, 1-55.
Hursh, D. (2007). Assessing No Child Left Behind and the rise of neoliberal education
policies. American Educational Research Journal, 44, 493-518.
Jackson, C. K., Johnson, R. C., & Persico, C. (2015). The effects of school spending on
educational and economic outcomes: Evidence from school finance reforms. The
Quarterly Journal of Economics, 131, 157-218.
Kline, R. B. (2010). Principles and Practice of Structural Equation Modeling, 3rd
Edition. The Guilford Press.
Kaiser, H. F. (1958). The varimax criterion for analytic rotation in factor analysis.
Psychometrika, 23, 187-200.
Leach, J. M., Scarborough, H. S., & Rescorla, L. (2003). Late-emerging reading
disabilities. Journal of educational psychology, 95, 211.
129
12
9
Pag
e12
9
2
2 P
age1
29
22
Pag
e12
92
2
Learning Point Associates. (2004). A closer look at the five essential components of
effective reading instruction: A review of scientifically based reading research for
teachers. Learning Point Associates.
Lee, J., Grigg, W., & Donahue, P. (2007). The nation's report card: Reading
2007 (NCES 2007–496). National Center for Education Statistics, Institute of
Education Sciences, U.S. Department of Education, Washington, D.C.
Lesaux, N. K., & Kieffer, M. J. (2010). Exploring sources of reading comprehension
difficulties among language minority learners and their classmates in early
adolescence. American Educational Research Journal, 47, 596-632.
Leslie, L., & Caldwell, J. S. (2006). Qualitative Reading Inventory (4th ed.). Boston,
MA: Allyn & Bacon.
Linn, R. L. (2000). Assessments and accountability. Educational Researcher, 29, 4-16.
Luchow, J. P., Crowl, T. K., & Kahn, J. P. (1985). Learned helplessness: Perceived
effects of ability and effort on academic performance among EH and LD/EH
children. Journal of learning disabilities, 18, 470-474.
Marinak, B. A., Malloy, J. B., Gambrell, L. B., & Mazzoni, S. A. (2015). Me and my
reading profile. The Reading Teacher, 69, 51-62.
Mazzoni, S. A., Gambrell, L. B., & Korkeamaki, R. L. (1999). A cross-cultural
perspective of early literacy motivation. Reading Psychology, 20, 237-253.
McCaslin, M. (2009). Co-regulation of student motivation and emergent
identity. Educational Psychologist, 44, 137-146.
130
13
0
Pag
e13
0
2
2 P
age1
30
22
Pag
e13
02
2
McDonald, R. P., & Ho, M. H. R. (2002). Principles and practice in reporting structural
equation analyses. Psychological methods, 7(1), 64.
McKenna, M.C., Kear, D.J., & Ellsworth, R.A. (1995). Children's attitudes toward
reading: A national survey. Reading Research Quarterly, 30, 934–955.
Milner, H. R. (2013). Rethinking achievement gap talk in urban education. Urban
Education, 48, 3-8.
Meyer, C. K., Trathen, W., Morris, D., McGee, J., Stewart, T. T., Vines, N. A., & Gill, T.
(2013) Reading profiles of struggling middle-school readers: What does it mean
in the era of the Common Core State Standards. Paper presented at the annual
meeting of the Literacy Researchers Association, Dallas, Texas. Abstract
retrieved from
https://secure.literacyresearchassociation.org/papers/lra2013123204531.doc
Moje, E.B. (2004). Federal adolescent literacy policy: Implications for administration,
policy, and the adolescent literacy research community. Paper presented at the
annual meeting of the National Reading Conference, San Antonio, TX.
National Center for Education Statistics. (2015). The nation's report card: Reading
2015 (NCES 2015–457). Washington, DC: Institute of Education Sciences & U.
S. Department of Education.
National Reading Panel. (2000). Report of the National Reading Panel teaching children
to read: An evidence-based assessment of the scientific research literature on
reading and its implications for reading instruction. Washington, DC: National
Institute of Child Health and Human Development.
131
13
1
Pag
e13
1
2
2 P
age1
31
22
Pag
e13
12
2
No Child Left Behind Act of 2001, Pub. L. No. 107-110.
Orfield, G., & Kornhaber, M. L. (Eds.) (2001). Raising standards or raising barriers?
Inequality and high-stakes testing in public education. New York: The Century
Foundation Press.
Paris, S., & Turner, J. (1994). Situated motivation. In P. Pintrich, D. Brown, & C.
Weinstein (Eds.), Student motivation, cognition, and learning: Essays in honor of
Wilbert J. McKeachie (pp. 213–237). Hillsdale, NJ: Lawrence Erlbaum.
Pierce, M. E., Katzir, T., Wolf, M., & Noam, G. G. (2007). Clusters of second and third
grade dysfluent urban readers. Reading and Writing, 20, 885-907.
Rencher, A. C. & Christensen, W. F. (2012). Methods of Multivariate Analysis (3rd ed.).
Hoboken, NJ: Wiley.
Rupp, A. A., & Lesaux, N. K. (2006). Meeting expectations? An empirical investigation
of a standards-based assessment of reading comprehension. Educational
Evaluation and Policy Analysis, 28, 315-333.
SAS Institute Inc. (2013). SAS 9.4 Help and documentation. Cary, NC
Schiefele, U. (1999). Interest and learning from text. Scientific Studies of Reading,
3, 257-279.
Schiefele, U., Schaffner, E., Möller, J., & Wigfield, A. (2012). Dimensions of reading
motivation and their relation to reading behavior and competence. Reading
Research Quarterly, 47, 427-463.
Sharma, S. (1996). Applied Multivariate Techniques. John Wiley, New York.
132
13
2
Pag
e13
2
2
2 P
age1
32
22
Pag
e13
22
2
Shepard, L. (2000). The role of classroom assessment in teaching and learning,
CSE Technical Report 517, Los Angeles, CA, National Center for
Research on Evaluation, Standards, and Student Testing.
Slavin, R. E., Cheung, A., Holmes, G. C., Madden, N. A., & Chamberlain, A. .
(2013). Effects of a data-driven district reform model on state assessment
outcomes. American Educational Research Journal, 50, 371-396.
Steiger, J. H. (2007). Understanding the limitations of global fit assessment in
structural equation modeling. Personality and Individual differences, 42, 893-898.
Stumpf, S. A., Brief, A. P., & Hartman, K. (1987). Self-efficacy expectations and
coping with career-related events. Journal of Vocational Behavior, 31, 91-108.
Swanson, C. B., & Barlage, J. (2006). Influence: A study of the factors shaping
education policy. Bethesda, MD: Editorial Projects in Education Research Center.
Timm, N. (2002). Applied multivariate statistics. Secaucus, NJ: Springer.
Unrau, N., & Schlackman, J. (2006). Motivation and its relationship with reading
achievement in an urban middle school. Journal of Educational Research, 100,
81-101.
Vansteenkiste, M., Lens, W., & Deci, E. L. (2006). Intrinsic versus extrinsic goal
contents in self-determination theory: Another look at the quality of academic
motivation. Educational Psychologist, 41, 19-31.
Valencia, S. W. (2011). Using assessment to improve teaching and learning. In. S.
J. Samuels & A. E. Farstrup (Eds.). What research has to say about reading
instruction (4th ed., pp.379-405). Newark, DE: International Reading Association.
133
13
3
Pag
e13
3
2
2 P
age1
33
22
Pag
e13
32
2
Valencia, S. W., & Buly, M. R. (2004). Behind test scores: What struggling
readers really need. Reading Teacher, 57, 520-533.
Wagner, R. K., Torgesen, J. T., & Rashotte, C. A. (1994). The Comprehensive
Test of Phonological Processing. Austin, TX: PRO-ED.
Watkins, M. W., & Coffey, D. Y. (2004). Reading Motivation: Multidimensional
and Indeterminate. Journal of Educational Psychology, 96, 110-118.
Wang, J. H., & Guthrie, J. T. (2004). Modeling the effects of intrinsic
motivation, extrinsic motivation, amount of reading, and past
achievement on text comprehension between U.S. and Chinese students.
Reading Research Quarterly, 39, 162-186.
Wigfield, A. (1997). Children's motivations for reading and reading
engagement. In J.T. Guthrie & A. Wigfield (Eds.), Reading engagement:
Motivating readers through integrated instruction (pp. 14–33). Newark, DE:
International Reading Association.
Wigfield, A., & Guthrie, J. T. (1995). Dimensions of children’s motivations for
.reading: An initial study (Reading Research Report No. 34). Athens, GA:
National Reading Research Center, Universities of Georgia and Maryland College
Park.
Wigfield, A., & Guthrie, J. T. (1997). Relations of children’s motivation for
reading to the amount and breadth of their reading. Journal of Educational
Psychology, 89, 430–432.
134
13
4
Pag
e13
4
2
2 P
age1
34
22
Pag
e13
42
2
Wigfield, A., Guthrie, J. T., Tonks, S., & Perencevich, K. C. (2004). Children's
motivation for reading: Domain specificity and instructional influences. The
Journal of Educational Research, 97, 299-310.
Yuan, K. H. (2005). Fit indices versus test statistics. Multivariate Behavioral
Research, 40, 115–148.
135
13
5
Pag
e13
5
2
2 P
age1
35
22
Pag
e13
52
2
APPENDIX A
MOTIVATION FOR READING QUESTIONNAIRE ITEMS
Students choose one of the following options for each statement below.
If the statement is very different from you, circle a 1.
If the statement is a little different from you, circle a 2.
If the statement is a little like you, circle a 3.
If the statement is a lot like you, circle a 4.
1. I like being the best at reading.
2. I like it when the questions in books make me think.
3. I read to improve my grades.
4. If the teacher discusses something interesting I might read more about it.
5. I like hard, challenging books.
6. I enjoy a long, involved story or fiction book.
7. I know that I will do well in reading next year.
8. If a book is interesting I don’t care how hard it is to read.
9. I try to get more answers right than my friends.
10. I have favorite subjects that I like to read about.
11. I visit the library often with my family.
12. I make pictures in my mind when I read.
136
13
6
Pag
e13
6
2
2 P
age1
36
22
Pag
e13
62
2
13. I don’t like reading something when the words are too difficult.
14. I enjoy reading books about people in different counties.
15. I am a good reader.
16. I usually learn difficult things by reading.
17. It is very important to me to be a good reader.
18. My parents often tell me what a good job I am doing in reading.
19. I read to learn new information about topics that interest me.
20. If the project is interesting, I can read difficult material.
21. I learn more from reading than most students in the class.
22. I read stories about fantasy and make believe.
23. I read because I have to.
24. I don’t like vocabulary questions.
25. I like to read about new things.
26. I often read to my brother or my sister.
27. In comparison to other activities I do, it is very important to me to be a
good reader.
28. I like having the teacher say I read well.
29. I read about my hobbies to learn more about them.
137
13
7
Pag
e13
7
2
2 P
age1
37
22
Pag
e13
72
2
30. I like mysteries.
31. My friends and I like to trade things to read.
32. Complicated stories are no fun to read.
33. I read a lot of adventure stories.
34. I do as little schoolwork as possible in reading.
35. I feel like I make friends with people in good books.
36. Finishing every reading assignment is very important to me.
37. My friends sometimes tell me I am a good reader.
38. Grades are a good way to see how well you are doing in reading.
39. I like to help my friends with their schoolwork in reading.
40. I don’t like it when there are too many people in the story.
41. I am willing to work hard to read better than my friends.
42. I sometimes read to my parents.
43. I like to get compliments for my reading.
44. It is important for me to see my name on a list of good readers.
45. I talk to my friends about what I am reading.
46. I always try to finish my reading on time.
138
13
8
Pag
e13
8
2
2 P
age1
38
22
Pag
e13
82
2
47. I am happy when someone recognizes my reading.
48. I like to tell my family about what I am reading.
49. I like being the only one who knows an answer in something we read.
50. I look forward to finding out my reading grade.
51. I always do my reading work exactly as the teacher wants it.
52. I like to finish my reading before other students.
53. My parents ask me about my reading grade.